fix: docx/email parsing, extractor LLM driver, and chunker alignment (#17144)
## Summary
Three groups of changes across the Go ingestion pipeline:
### 1. DOCX parsing improvements
- **docx_parser.go**: Enhanced DOCX parsing with better structure
extraction and media handling
- **docx_parser_cgo_test.go**, **docx_parser_test.go**: Companion tests
- **office_parsers_no_cgo.go**: Stub sync for non-CGO builds
### 2. Email (.eml) parsing: base64 Content-Transfer-Encoding decoding
- **email_parser.go** (`decodeCTE`): Added Content-Transfer-Encoding
decoding for base64 and quoted-printable. Go's `mime/multipart.Reader`
does not decode Content-Transfer-Encoding automatically, so attachments
with `Content-Transfer-Encoding: base64` remained base64-encoded in the
output. The new `decodeCTE` helper is called after reading each
multipart part's raw bytes in `readMailBody`, mirroring Python's
`part.get_payload(decode=True)`.
- **email_parser_test.go**: Two new tests — simple base64 attachment and
nested multipart/alternative with base64 attachment.
### 3. Extractor LLM driver fix + ModelDriver consolidation
- **extractor.go**: Fixed a bug where the Extractor component used
`ModelFactory.CreateModelDriver()`, which creates bare model instances
without API keys or provider configuration. Switched to
`models.GetPreconfiguredDriver()` which resolves the actual
pre-configured driver from `ProviderManager`, matching the codepath used
by `llm.go`. This fixes auto keyword/question extraction in DSL
pipelines that require LLM calls.
- **get_driver.go** (new): Extracted shared `GetPreconfiguredDriver()`
from `llm.go:newChatModelDriver()` so both `llm.go` and `extractor.go`
use the same codepath.
- **get_driver_test.go** (new): Tests for the shared driver resolution.
- **llm.go**: Replaced inline driver resolution with
`models.GetPreconfiguredDriver()`.
### 4. Chunker fixes and observability
- **group.go** (`extractLineRecords`): Fixed to also read `markdown` and
`html` payload keys — previously it only read `text`/`content`, causing
GroupTitleChunker to silently return empty results for markdown-format
parser output.
- **common.go** (`compileDelimPattern`): Aligned with Python's
`_compile_delimiter_pattern` — only backtick-wrapped delimiters produce
an active regex pattern; plain delimiters are not compiled into the
split regex.
- **token.go** (`applyChildrenDelim`): Set `DocType` and `CKType` to
`"text"` on created ChunkDocs so the token-size merge path correctly
identifies and merges text segments.
- **parser.go**, **extractor.go**, **tokenizer.go**, **group.go**,
**hierarchy.go**: Added debug-level logging for pipeline diagnostics.
- **parser_dispatch_test.go**, **group_test.go**: New tests.
## Verification
- All Go tests pass: `bash build.sh --test ./internal/parser/parser/...`
and `bash build.sh --test ./internal/ingestion/component/...`
- Build succeeds: `bash build.sh --go`
2026-07-21 13:51:17 +08:00
|
|
|
//go:build cgo
|
|
|
|
|
|
|
|
|
|
//
|
|
|
|
|
// Copyright 2026 The InfiniFlow Authors. All Rights Reserved.
|
|
|
|
|
//
|
|
|
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
|
// you may not use this file except in compliance with the License.
|
|
|
|
|
// You may obtain a copy of the License at
|
|
|
|
|
//
|
|
|
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
|
//
|
|
|
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
|
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
|
// See the License for the specific language governing permissions and
|
|
|
|
|
// limitations under the License.
|
|
|
|
|
//
|
|
|
|
|
|
|
|
|
|
package parser
|
|
|
|
|
|
|
|
|
|
import (
|
|
|
|
|
"encoding/base64"
|
fix(ingestion): align laws DSL with Python — heading fallback, colon-title, short-line filter, remove_toc, and image extension mapping (#17200)
## Summary
This PR aligns the Go ingestion pipeline's **Laws** DSL template with
the Python implementation by fixing heading-detection gaps, adds
image-extension support, refactors the **Extractor** component's LLM
resolution, hardens heading detection for CJK text, and makes the
Extractor accept the Python DSL prompt key names
(`sys_prompt`/`prompts`) alongside the Go names.
## Changes
### 1. Picture file-type detection (`internal/utility/file.go`)
Adds explicit mapping for common image extensions (png, jpg, jpeg, gif,
bmp, tiff, tif, webp, svg, ico, avif, heic, apng) → `FileTypeVISUAL`,
with regression tests.
### 2. Laws DSL heading-detection alignment
(`internal/ingestion/component/chunker/`)
Four fixes to `resolveTitleLevels`:
| Fix | What changed | Why |
|-----|-------------|-----|
| **DOCX `ck_type` fallback** | `ckType` field on `lineRecord`,
propagated from `ChunkDoc.CKType` in `recordsFromStructured`. When
`ck_type=="heading"`, assign `fallbackLevel`. | office_oxide extracts
DOCX heading metadata, but the info was lost before reaching the heading
detector. Word headings whose text doesn't match any regex (e.g.
"Introduction") were treated as body. |
| **`make_colon_as_title` promotion** | `isColonTitle()`: promotes lines
ending with `:`/`:` that have sentence-ending punctuation before the
colon and ≥32 runes between them. | Mirrors Python's
`make_colon_as_title` in `rag/nlp/__init__.py`. Triple guard prevents
false positives. |
| **Short/numeric line filter** | Lines with ≤1 rune or purely numeric
are pinned to body level. | Mirrors Python `tree_merge`'s filter of
`sections` where `len(...) <= 1` or `re.match(r"[0-9]+$", ...)`. |
| **PDF `remove_toc`** | `"remove_toc": true` added to the PDF parser
setup in `ingestion_pipeline_laws.json`. | The Go PDF parser already
supports TOC removal; the Book template already enables it. |
### 3. Extractor llm_id resolution
(`internal/ingestion/component/extractor.go`)
Refactored to handle both **bare tenant_model UUIDs** and **composite
model@provider** strings via the shared `resolveModelConfig`
(`dispatch_model.go`):
- **`resolveExtractorChatConfig`** — UUID path calls
`resolveModelConfigByID` directly (one DB hit); composite path goes
through `resolveModelConfig`. Added `isBareTenantModelID` pre-check for
clear errors when a UUID doesn't exist.
- **`resolveExtractorChatTarget`** — propagates resolution errors
instead of silently returning empty driver.
- **`Chat()`** — removed `driver = "dummy"` fallback. Missing driver is
now an explicit error.
- **Removed dead code**: `splitExtractorLLID`,
`findExtractorSoleActiveInstance`.
### 4. `InjectExtractorLLMID` — fallback when no user config
(`internal/common/parser_config.go`)
Injects the tenant's global default LLM into extractor components **only
when their `llm_id` is empty**. Preserves user-selected UUID or
model@provider values.
Priority: user-configured llm_id > tenant global default > error (no
silent dummy fallback).
### 5. `ResponseHeaderTimeout` increase
(`internal/entity/models/base_model.go`)
`ResponseHeaderTimeout` 60s → 120s in `NewDriverHTTPClient`. Reasoning
models with large extraction prompts can take longer than 60s to produce
the first response token.
### 6. CJK rune-aware heading detection
(`internal/ingestion/component/chunker/title.go`)
Two byte-vs-rune bugs that only manifest on CJK text:
| Fix | What changed | Why |
|-----|-------------|-----|
| **`isColonTitle` byte offset** | `body[lastPunct+1:]` →
`body[lastPunct+runeLen:]` via `utf8.DecodeRuneInString` |
`strings.LastIndexAny` returns a byte index; `+1` skips only 1 byte,
corrupting multi-byte CJK punctuation (e.g. `。` = 3 bytes) and inflating
the rune count past the 32-rune threshold → false-positive heading
promotion. |
| **Short-line filter byte count** | `len(text) <= 1` →
`utf8.RuneCountInString(text) <= 1` | Go `len` is UTF-8 bytes; a single
CJK char (3 bytes) passed the filter, but Python's `len` returns 1 →
mismatch. |
### 7. `extractor_tag.go` — log error when llm fails
When `resolveExtractorChatTarget` returned an error, `runAutoTags` will
log error.
### 8. Python DSL prompt-key compatibility
(`internal/ingestion/component/extractor.go`)
The Resume DSL template uses Python-side key names (`sys_prompt`,
`prompts`). `NewExtractorComponent` now accepts them as fallbacks
alongside the Go names:
- `system_prompt` (Go) ← `sys_prompt` (Python) as fallback
- `prompt` (Go string) ← `prompts` (Python array `[{"role","content"}]`,
takes `[0].content`) as fallback
Mirrors the alias pattern already in `internal/agent/component/llm.go`.
`resolveInputs` accepts per-call `sys_prompt` override too.
## Remaining gaps vs Python
| Gap | Scope | Impact |
|-----|-------|--------|
| **TOC removal for TXT/MD/HTML** | Python's `remove_contents_table`
works on all text formats; Go's `remove_toc` is PDF-only. | Low —
plain-text documents rarely contain structured TOCs. |
| **Regex pattern details** | Minor differences in quantifiers, missing
H5/H6 markdown patterns, missing 4-level numbering pattern. | Low — Go's
variants are stricter; DOCX headings are covered by `ck_type` fallback.
|
## Testing
- `TestHierarchyTitleChunker_CKTypeHeadingFallback` — DOCX `ck_type`
heading promotion
- `TestHierarchyTitleChunker_ColonTitlePromotion` /
`_ColonTitleShortLine_Negative` — colon-title promotion + guard
- `TestIsColonTitle_CJKEdgeCase` / `TestIsColonTitle_ASCII_NoRegression`
— CJK byte-offset fix + ASCII regression
- `TestHierarchyTitleChunker_ColonTitlePromotion_CJK_EdgeCase` — CJK
colon edge case through full pipeline
- `TestHierarchyTitleChunker_ShortSingleCJKLineFilter` — single CJK char
filtered to body
- `TestHierarchyTitleChunker_ShortNumericLineFilter` — purely numeric
lines filtered
- `TestGetFileType_ImageExtensions` / `_ExistingFormats_NoRegression` —
image extension mapping
- `TestInjectExtractorLLMID_SkipWhenUUID` / `_SkipWhenComposite` /
`_InjectWhenEmpty` — llm_id injection guard
- `TestIsBareTenantModelID` — UUID detection
- `TestResolveExtractorChatTarget_AtSplitFallback` / `_NoDriver` — @
split fallback without DB
- `TestNewExtractorComponent_SysPromptAlias` / `_PromptsArray` /
`_PromptsArray_PromptWins` / `_SystemPromptWinsOverSysPrompt` — Python
key compatibility
- `TestBuildDOCXJSONSections_List` / `_TextBox` / `_MixedWithList` —
DOCX list/text_box parsing
- Full ingestion test suite passes (chunker, pipeline, task, service,
component packages)
2026-07-22 19:14:32 +08:00
|
|
|
"encoding/json"
|
fix: docx/email parsing, extractor LLM driver, and chunker alignment (#17144)
## Summary
Three groups of changes across the Go ingestion pipeline:
### 1. DOCX parsing improvements
- **docx_parser.go**: Enhanced DOCX parsing with better structure
extraction and media handling
- **docx_parser_cgo_test.go**, **docx_parser_test.go**: Companion tests
- **office_parsers_no_cgo.go**: Stub sync for non-CGO builds
### 2. Email (.eml) parsing: base64 Content-Transfer-Encoding decoding
- **email_parser.go** (`decodeCTE`): Added Content-Transfer-Encoding
decoding for base64 and quoted-printable. Go's `mime/multipart.Reader`
does not decode Content-Transfer-Encoding automatically, so attachments
with `Content-Transfer-Encoding: base64` remained base64-encoded in the
output. The new `decodeCTE` helper is called after reading each
multipart part's raw bytes in `readMailBody`, mirroring Python's
`part.get_payload(decode=True)`.
- **email_parser_test.go**: Two new tests — simple base64 attachment and
nested multipart/alternative with base64 attachment.
### 3. Extractor LLM driver fix + ModelDriver consolidation
- **extractor.go**: Fixed a bug where the Extractor component used
`ModelFactory.CreateModelDriver()`, which creates bare model instances
without API keys or provider configuration. Switched to
`models.GetPreconfiguredDriver()` which resolves the actual
pre-configured driver from `ProviderManager`, matching the codepath used
by `llm.go`. This fixes auto keyword/question extraction in DSL
pipelines that require LLM calls.
- **get_driver.go** (new): Extracted shared `GetPreconfiguredDriver()`
from `llm.go:newChatModelDriver()` so both `llm.go` and `extractor.go`
use the same codepath.
- **get_driver_test.go** (new): Tests for the shared driver resolution.
- **llm.go**: Replaced inline driver resolution with
`models.GetPreconfiguredDriver()`.
### 4. Chunker fixes and observability
- **group.go** (`extractLineRecords`): Fixed to also read `markdown` and
`html` payload keys — previously it only read `text`/`content`, causing
GroupTitleChunker to silently return empty results for markdown-format
parser output.
- **common.go** (`compileDelimPattern`): Aligned with Python's
`_compile_delimiter_pattern` — only backtick-wrapped delimiters produce
an active regex pattern; plain delimiters are not compiled into the
split regex.
- **token.go** (`applyChildrenDelim`): Set `DocType` and `CKType` to
`"text"` on created ChunkDocs so the token-size merge path correctly
identifies and merges text segments.
- **parser.go**, **extractor.go**, **tokenizer.go**, **group.go**,
**hierarchy.go**: Added debug-level logging for pipeline diagnostics.
- **parser_dispatch_test.go**, **group_test.go**: New tests.
## Verification
- All Go tests pass: `bash build.sh --test ./internal/parser/parser/...`
and `bash build.sh --test ./internal/ingestion/component/...`
- Build succeeds: `bash build.sh --go`
2026-07-21 13:51:17 +08:00
|
|
|
"strings"
|
|
|
|
|
"testing"
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_Paragraphs(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"paragraph","content":[{"type":"text","text":"Hello world"}],"style":"Normal"}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 1 {
|
|
|
|
|
t.Fatalf("got %d sections, want 1", len(sections))
|
|
|
|
|
}
|
|
|
|
|
item := sections[0]
|
|
|
|
|
if got, ok := item["text"].(string); !ok || got != "Hello world" {
|
|
|
|
|
t.Errorf("text = %q, want %q", got, "Hello world")
|
|
|
|
|
}
|
|
|
|
|
if got, ok := item["doc_type_kwd"].(string); !ok || got != "text" {
|
|
|
|
|
t.Errorf("doc_type_kwd = %q, want %q", got, "text")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_Headings(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"heading","level":1,"content":[{"type":"text","text":"Chapter 1"}]}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 1 {
|
|
|
|
|
t.Fatalf("got %d sections, want 1", len(sections))
|
|
|
|
|
}
|
|
|
|
|
item := sections[0]
|
|
|
|
|
if got, ok := item["text"].(string); !ok || got != "Chapter 1" {
|
|
|
|
|
t.Errorf("text = %q, want %q", got, "Chapter 1")
|
|
|
|
|
}
|
|
|
|
|
if got, ok := item["doc_type_kwd"].(string); !ok || got != "text" {
|
|
|
|
|
t.Errorf("doc_type_kwd = %q, want %q", got, "text")
|
|
|
|
|
}
|
|
|
|
|
if got, ok := item["ck_type"].(string); !ok || got != "heading" {
|
|
|
|
|
t.Errorf("ck_type = %q, want %q", got, "heading")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_Images(t *testing.T) {
|
|
|
|
|
b64 := base64.StdEncoding.EncodeToString([]byte("fake-image-data"))
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"image","data":"` + b64 + `"}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 1 {
|
|
|
|
|
t.Fatalf("got %d sections, want 1", len(sections))
|
|
|
|
|
}
|
|
|
|
|
item := sections[0]
|
|
|
|
|
if got, ok := item["text"].(string); !ok || got != "" {
|
|
|
|
|
t.Errorf("text = %q, want empty", got)
|
|
|
|
|
}
|
|
|
|
|
if got, ok := item["image"].(string); !ok || got != b64 {
|
|
|
|
|
t.Errorf("image = %q, want %q", got, b64)
|
|
|
|
|
}
|
|
|
|
|
if got, ok := item["doc_type_kwd"].(string); !ok || got != "image" {
|
|
|
|
|
t.Errorf("doc_type_kwd = %q, want %q", got, "image")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_Tables(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"table","rows":[
|
|
|
|
|
{"cells":[{"content":[{"type":"paragraph","content":[{"type":"text","text":"A1"}]}]},{"content":[{"type":"paragraph","content":[{"type":"text","text":"B1"}]}]}]},
|
|
|
|
|
{"cells":[{"content":[{"type":"paragraph","content":[{"type":"text","text":"A2"}]}]},{"content":[{"type":"paragraph","content":[{"type":"text","text":"B2"}]}]}]}
|
|
|
|
|
]}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 1 {
|
|
|
|
|
t.Fatalf("got %d sections, want 1", len(sections))
|
|
|
|
|
}
|
|
|
|
|
item := sections[0]
|
|
|
|
|
html, ok := item["text"].(string)
|
|
|
|
|
if !ok {
|
|
|
|
|
t.Fatal("text field missing or not string")
|
|
|
|
|
}
|
|
|
|
|
if !strings.Contains(html, "<table>") || !strings.Contains(html, "</table>") {
|
|
|
|
|
t.Errorf("html = %q, missing <table> tags", html)
|
|
|
|
|
}
|
|
|
|
|
if !strings.Contains(html, "<tr>") || !strings.Contains(html, "</tr>") {
|
|
|
|
|
t.Errorf("html = %q, missing <tr> tags", html)
|
|
|
|
|
}
|
|
|
|
|
if !strings.Contains(html, "<td>A1</td>") || !strings.Contains(html, "<td>B2</td>") {
|
|
|
|
|
t.Errorf("html = %q, missing cell content", html)
|
|
|
|
|
}
|
|
|
|
|
if got, ok := item["doc_type_kwd"].(string); !ok || got != "table" {
|
|
|
|
|
t.Errorf("doc_type_kwd = %q, want %q", got, "table")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_MixedContent(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"paragraph","content":[{"type":"text","text":"First para"}]},
|
|
|
|
|
{"type":"image","data":"aW1hZ2U="},
|
|
|
|
|
{"type":"table","rows":[{"cells":[{"content":[{"type":"paragraph","content":[{"type":"text","text":"cell"}]}]}]}]},
|
|
|
|
|
{"type":"heading","level":2,"content":[{"type":"text","text":"Sub title"}]}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 4 {
|
|
|
|
|
t.Fatalf("got %d sections, want 4", len(sections))
|
|
|
|
|
}
|
|
|
|
|
// paragraph first
|
|
|
|
|
if got, _ := sections[0]["doc_type_kwd"].(string); got != "text" {
|
|
|
|
|
t.Errorf("item[0].doc_type_kwd = %q, want %q", got, "text")
|
|
|
|
|
}
|
|
|
|
|
// image second
|
|
|
|
|
if got, _ := sections[1]["doc_type_kwd"].(string); got != "image" {
|
|
|
|
|
t.Errorf("item[1].doc_type_kwd = %q, want %q", got, "image")
|
|
|
|
|
}
|
|
|
|
|
// table third
|
|
|
|
|
if got, _ := sections[2]["doc_type_kwd"].(string); got != "table" {
|
|
|
|
|
t.Errorf("item[2].doc_type_kwd = %q, want %q", got, "table")
|
|
|
|
|
}
|
|
|
|
|
// heading fourth
|
|
|
|
|
if got, _ := sections[3]["doc_type_kwd"].(string); got != "text" {
|
|
|
|
|
t.Errorf("item[3].doc_type_kwd = %q, want %q", got, "text")
|
|
|
|
|
}
|
|
|
|
|
if got, _ := sections[3]["ck_type"].(string); got != "heading" {
|
|
|
|
|
t.Errorf("item[3].ck_type = %q, want %q", got, "heading")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
fix(ingestion): align laws DSL with Python — heading fallback, colon-title, short-line filter, remove_toc, and image extension mapping (#17200)
## Summary
This PR aligns the Go ingestion pipeline's **Laws** DSL template with
the Python implementation by fixing heading-detection gaps, adds
image-extension support, refactors the **Extractor** component's LLM
resolution, hardens heading detection for CJK text, and makes the
Extractor accept the Python DSL prompt key names
(`sys_prompt`/`prompts`) alongside the Go names.
## Changes
### 1. Picture file-type detection (`internal/utility/file.go`)
Adds explicit mapping for common image extensions (png, jpg, jpeg, gif,
bmp, tiff, tif, webp, svg, ico, avif, heic, apng) → `FileTypeVISUAL`,
with regression tests.
### 2. Laws DSL heading-detection alignment
(`internal/ingestion/component/chunker/`)
Four fixes to `resolveTitleLevels`:
| Fix | What changed | Why |
|-----|-------------|-----|
| **DOCX `ck_type` fallback** | `ckType` field on `lineRecord`,
propagated from `ChunkDoc.CKType` in `recordsFromStructured`. When
`ck_type=="heading"`, assign `fallbackLevel`. | office_oxide extracts
DOCX heading metadata, but the info was lost before reaching the heading
detector. Word headings whose text doesn't match any regex (e.g.
"Introduction") were treated as body. |
| **`make_colon_as_title` promotion** | `isColonTitle()`: promotes lines
ending with `:`/`:` that have sentence-ending punctuation before the
colon and ≥32 runes between them. | Mirrors Python's
`make_colon_as_title` in `rag/nlp/__init__.py`. Triple guard prevents
false positives. |
| **Short/numeric line filter** | Lines with ≤1 rune or purely numeric
are pinned to body level. | Mirrors Python `tree_merge`'s filter of
`sections` where `len(...) <= 1` or `re.match(r"[0-9]+$", ...)`. |
| **PDF `remove_toc`** | `"remove_toc": true` added to the PDF parser
setup in `ingestion_pipeline_laws.json`. | The Go PDF parser already
supports TOC removal; the Book template already enables it. |
### 3. Extractor llm_id resolution
(`internal/ingestion/component/extractor.go`)
Refactored to handle both **bare tenant_model UUIDs** and **composite
model@provider** strings via the shared `resolveModelConfig`
(`dispatch_model.go`):
- **`resolveExtractorChatConfig`** — UUID path calls
`resolveModelConfigByID` directly (one DB hit); composite path goes
through `resolveModelConfig`. Added `isBareTenantModelID` pre-check for
clear errors when a UUID doesn't exist.
- **`resolveExtractorChatTarget`** — propagates resolution errors
instead of silently returning empty driver.
- **`Chat()`** — removed `driver = "dummy"` fallback. Missing driver is
now an explicit error.
- **Removed dead code**: `splitExtractorLLID`,
`findExtractorSoleActiveInstance`.
### 4. `InjectExtractorLLMID` — fallback when no user config
(`internal/common/parser_config.go`)
Injects the tenant's global default LLM into extractor components **only
when their `llm_id` is empty**. Preserves user-selected UUID or
model@provider values.
Priority: user-configured llm_id > tenant global default > error (no
silent dummy fallback).
### 5. `ResponseHeaderTimeout` increase
(`internal/entity/models/base_model.go`)
`ResponseHeaderTimeout` 60s → 120s in `NewDriverHTTPClient`. Reasoning
models with large extraction prompts can take longer than 60s to produce
the first response token.
### 6. CJK rune-aware heading detection
(`internal/ingestion/component/chunker/title.go`)
Two byte-vs-rune bugs that only manifest on CJK text:
| Fix | What changed | Why |
|-----|-------------|-----|
| **`isColonTitle` byte offset** | `body[lastPunct+1:]` →
`body[lastPunct+runeLen:]` via `utf8.DecodeRuneInString` |
`strings.LastIndexAny` returns a byte index; `+1` skips only 1 byte,
corrupting multi-byte CJK punctuation (e.g. `。` = 3 bytes) and inflating
the rune count past the 32-rune threshold → false-positive heading
promotion. |
| **Short-line filter byte count** | `len(text) <= 1` →
`utf8.RuneCountInString(text) <= 1` | Go `len` is UTF-8 bytes; a single
CJK char (3 bytes) passed the filter, but Python's `len` returns 1 →
mismatch. |
### 7. `extractor_tag.go` — log error when llm fails
When `resolveExtractorChatTarget` returned an error, `runAutoTags` will
log error.
### 8. Python DSL prompt-key compatibility
(`internal/ingestion/component/extractor.go`)
The Resume DSL template uses Python-side key names (`sys_prompt`,
`prompts`). `NewExtractorComponent` now accepts them as fallbacks
alongside the Go names:
- `system_prompt` (Go) ← `sys_prompt` (Python) as fallback
- `prompt` (Go string) ← `prompts` (Python array `[{"role","content"}]`,
takes `[0].content`) as fallback
Mirrors the alias pattern already in `internal/agent/component/llm.go`.
`resolveInputs` accepts per-call `sys_prompt` override too.
## Remaining gaps vs Python
| Gap | Scope | Impact |
|-----|-------|--------|
| **TOC removal for TXT/MD/HTML** | Python's `remove_contents_table`
works on all text formats; Go's `remove_toc` is PDF-only. | Low —
plain-text documents rarely contain structured TOCs. |
| **Regex pattern details** | Minor differences in quantifiers, missing
H5/H6 markdown patterns, missing 4-level numbering pattern. | Low — Go's
variants are stricter; DOCX headings are covered by `ck_type` fallback.
|
## Testing
- `TestHierarchyTitleChunker_CKTypeHeadingFallback` — DOCX `ck_type`
heading promotion
- `TestHierarchyTitleChunker_ColonTitlePromotion` /
`_ColonTitleShortLine_Negative` — colon-title promotion + guard
- `TestIsColonTitle_CJKEdgeCase` / `TestIsColonTitle_ASCII_NoRegression`
— CJK byte-offset fix + ASCII regression
- `TestHierarchyTitleChunker_ColonTitlePromotion_CJK_EdgeCase` — CJK
colon edge case through full pipeline
- `TestHierarchyTitleChunker_ShortSingleCJKLineFilter` — single CJK char
filtered to body
- `TestHierarchyTitleChunker_ShortNumericLineFilter` — purely numeric
lines filtered
- `TestGetFileType_ImageExtensions` / `_ExistingFormats_NoRegression` —
image extension mapping
- `TestInjectExtractorLLMID_SkipWhenUUID` / `_SkipWhenComposite` /
`_InjectWhenEmpty` — llm_id injection guard
- `TestIsBareTenantModelID` — UUID detection
- `TestResolveExtractorChatTarget_AtSplitFallback` / `_NoDriver` — @
split fallback without DB
- `TestNewExtractorComponent_SysPromptAlias` / `_PromptsArray` /
`_PromptsArray_PromptWins` / `_SystemPromptWinsOverSysPrompt` — Python
key compatibility
- `TestBuildDOCXJSONSections_List` / `_TextBox` / `_MixedWithList` —
DOCX list/text_box parsing
- Full ingestion test suite passes (chunker, pipeline, task, service,
component packages)
2026-07-22 19:14:32 +08:00
|
|
|
func TestBuildDOCXJSONSections_List(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"list","ordered":false,"items":[
|
|
|
|
|
{"content":[{"type":"paragraph","content":[{"type":"text","text":"Item 1"}]}]},
|
|
|
|
|
{"content":[{"type":"paragraph","content":[{"type":"text","text":"Item 2"}]}]},
|
|
|
|
|
{"content":[{"type":"paragraph","content":[{"type":"text","text":"Item 3"}]}]}
|
|
|
|
|
],"level":0}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 3 {
|
|
|
|
|
t.Fatalf("got %d sections, want 3 (each list item should be a section)", len(sections))
|
|
|
|
|
}
|
|
|
|
|
for i, want := range []string{"Item 1", "Item 2", "Item 3"} {
|
|
|
|
|
if got, _ := sections[i]["text"].(string); got != want {
|
|
|
|
|
t.Errorf("section[%d].text = %q, want %q", i, got, want)
|
|
|
|
|
}
|
|
|
|
|
if got, ok := sections[i]["doc_type_kwd"].(string); !ok || got != "text" {
|
|
|
|
|
t.Errorf("section[%d].doc_type_kwd = %q, want %q", i, got, "text")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_TextBox(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"text_box","content":[{"type":"paragraph","content":[{"type":"text","text":"Boxed paragraph"}]}],"width_emu":null}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 1 {
|
|
|
|
|
t.Fatalf("got %d sections, want 1 (text_box content should become a section)", len(sections))
|
|
|
|
|
}
|
|
|
|
|
if got, _ := sections[0]["text"].(string); got != "Boxed paragraph" {
|
|
|
|
|
t.Errorf("text = %q, want %q", got, "Boxed paragraph")
|
|
|
|
|
}
|
|
|
|
|
if got, ok := sections[0]["doc_type_kwd"].(string); !ok || got != "text" {
|
|
|
|
|
t.Errorf("doc_type_kwd = %q, want %q", got, "text")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestBuildDOCXJSONSections_MixedWithList(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"paragraph","content":[{"type":"text","text":"Preamble"}],"style":"Normal"},
|
|
|
|
|
{"type":"list","ordered":false,"items":[
|
|
|
|
|
{"content":[{"type":"paragraph","content":[{"type":"text","text":"Bullet A"}]}]},
|
|
|
|
|
{"content":[{"type":"paragraph","content":[{"type":"text","text":"Bullet B"}]}]}
|
|
|
|
|
],"level":0},
|
|
|
|
|
{"type":"paragraph","content":[{"type":"text","text":"Trailer"}],"style":"Normal"}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 4 {
|
|
|
|
|
t.Fatalf("got %d sections, want 4 (para + 2 list items + para)", len(sections))
|
|
|
|
|
}
|
|
|
|
|
wants := []string{"Preamble", "Bullet A", "Bullet B", "Trailer"}
|
|
|
|
|
for i, want := range wants {
|
|
|
|
|
if got, _ := sections[i]["text"].(string); got != want {
|
|
|
|
|
t.Errorf("section[%d].text = %q, want %q", i, got, want)
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
fix: docx/email parsing, extractor LLM driver, and chunker alignment (#17144)
## Summary
Three groups of changes across the Go ingestion pipeline:
### 1. DOCX parsing improvements
- **docx_parser.go**: Enhanced DOCX parsing with better structure
extraction and media handling
- **docx_parser_cgo_test.go**, **docx_parser_test.go**: Companion tests
- **office_parsers_no_cgo.go**: Stub sync for non-CGO builds
### 2. Email (.eml) parsing: base64 Content-Transfer-Encoding decoding
- **email_parser.go** (`decodeCTE`): Added Content-Transfer-Encoding
decoding for base64 and quoted-printable. Go's `mime/multipart.Reader`
does not decode Content-Transfer-Encoding automatically, so attachments
with `Content-Transfer-Encoding: base64` remained base64-encoded in the
output. The new `decodeCTE` helper is called after reading each
multipart part's raw bytes in `readMailBody`, mirroring Python's
`part.get_payload(decode=True)`.
- **email_parser_test.go**: Two new tests — simple base64 attachment and
nested multipart/alternative with base64 attachment.
### 3. Extractor LLM driver fix + ModelDriver consolidation
- **extractor.go**: Fixed a bug where the Extractor component used
`ModelFactory.CreateModelDriver()`, which creates bare model instances
without API keys or provider configuration. Switched to
`models.GetPreconfiguredDriver()` which resolves the actual
pre-configured driver from `ProviderManager`, matching the codepath used
by `llm.go`. This fixes auto keyword/question extraction in DSL
pipelines that require LLM calls.
- **get_driver.go** (new): Extracted shared `GetPreconfiguredDriver()`
from `llm.go:newChatModelDriver()` so both `llm.go` and `extractor.go`
use the same codepath.
- **get_driver_test.go** (new): Tests for the shared driver resolution.
- **llm.go**: Replaced inline driver resolution with
`models.GetPreconfiguredDriver()`.
### 4. Chunker fixes and observability
- **group.go** (`extractLineRecords`): Fixed to also read `markdown` and
`html` payload keys — previously it only read `text`/`content`, causing
GroupTitleChunker to silently return empty results for markdown-format
parser output.
- **common.go** (`compileDelimPattern`): Aligned with Python's
`_compile_delimiter_pattern` — only backtick-wrapped delimiters produce
an active regex pattern; plain delimiters are not compiled into the
split regex.
- **token.go** (`applyChildrenDelim`): Set `DocType` and `CKType` to
`"text"` on created ChunkDocs so the token-size merge path correctly
identifies and merges text segments.
- **parser.go**, **extractor.go**, **tokenizer.go**, **group.go**,
**hierarchy.go**: Added debug-level logging for pipeline diagnostics.
- **parser_dispatch_test.go**, **group_test.go**: New tests.
## Verification
- All Go tests pass: `bash build.sh --test ./internal/parser/parser/...`
and `bash build.sh --test ./internal/ingestion/component/...`
- Build succeeds: `bash build.sh --go`
2026-07-21 13:51:17 +08:00
|
|
|
func TestBuildDOCXJSONSections_EmptySkipped(t *testing.T) {
|
|
|
|
|
ir := `{"sections":[{"title":"","elements":[
|
|
|
|
|
{"type":"paragraph","content":[{"type":"text","text":""}]},
|
|
|
|
|
{"type":"paragraph","content":[{"type":"text","text":"Only this matters"}]},
|
|
|
|
|
{"type":"table","rows":[]}
|
|
|
|
|
]}]}`
|
|
|
|
|
sections := buildDOCXJSONSections(ir)
|
|
|
|
|
if len(sections) != 1 {
|
|
|
|
|
t.Fatalf("got %d sections, want 1 (only non-empty paragraph)", len(sections))
|
|
|
|
|
}
|
|
|
|
|
if got, _ := sections[0]["text"].(string); got != "Only this matters" {
|
|
|
|
|
t.Errorf("text = %q, want %q", got, "Only this matters")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestDocxIRTableToHTML_Empty(t *testing.T) {
|
|
|
|
|
el := docxIRElement{Type: "table", Rows: []docxIRRow{}}
|
|
|
|
|
html := docxIRTableToHTML(el)
|
|
|
|
|
if html != "<table></table>" {
|
|
|
|
|
t.Errorf("empty table html = %q, want %q", html, "<table></table>")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestDocxIRTableToHTML_Single(t *testing.T) {
|
|
|
|
|
el := docxIRElement{
|
|
|
|
|
Type: "table",
|
|
|
|
|
Rows: []docxIRRow{{
|
|
|
|
|
Cells: []docxIRCell{{
|
|
|
|
|
Content: []docxIRElement{{
|
|
|
|
|
Type: "paragraph",
|
fix(ingestion): align laws DSL with Python — heading fallback, colon-title, short-line filter, remove_toc, and image extension mapping (#17200)
## Summary
This PR aligns the Go ingestion pipeline's **Laws** DSL template with
the Python implementation by fixing heading-detection gaps, adds
image-extension support, refactors the **Extractor** component's LLM
resolution, hardens heading detection for CJK text, and makes the
Extractor accept the Python DSL prompt key names
(`sys_prompt`/`prompts`) alongside the Go names.
## Changes
### 1. Picture file-type detection (`internal/utility/file.go`)
Adds explicit mapping for common image extensions (png, jpg, jpeg, gif,
bmp, tiff, tif, webp, svg, ico, avif, heic, apng) → `FileTypeVISUAL`,
with regression tests.
### 2. Laws DSL heading-detection alignment
(`internal/ingestion/component/chunker/`)
Four fixes to `resolveTitleLevels`:
| Fix | What changed | Why |
|-----|-------------|-----|
| **DOCX `ck_type` fallback** | `ckType` field on `lineRecord`,
propagated from `ChunkDoc.CKType` in `recordsFromStructured`. When
`ck_type=="heading"`, assign `fallbackLevel`. | office_oxide extracts
DOCX heading metadata, but the info was lost before reaching the heading
detector. Word headings whose text doesn't match any regex (e.g.
"Introduction") were treated as body. |
| **`make_colon_as_title` promotion** | `isColonTitle()`: promotes lines
ending with `:`/`:` that have sentence-ending punctuation before the
colon and ≥32 runes between them. | Mirrors Python's
`make_colon_as_title` in `rag/nlp/__init__.py`. Triple guard prevents
false positives. |
| **Short/numeric line filter** | Lines with ≤1 rune or purely numeric
are pinned to body level. | Mirrors Python `tree_merge`'s filter of
`sections` where `len(...) <= 1` or `re.match(r"[0-9]+$", ...)`. |
| **PDF `remove_toc`** | `"remove_toc": true` added to the PDF parser
setup in `ingestion_pipeline_laws.json`. | The Go PDF parser already
supports TOC removal; the Book template already enables it. |
### 3. Extractor llm_id resolution
(`internal/ingestion/component/extractor.go`)
Refactored to handle both **bare tenant_model UUIDs** and **composite
model@provider** strings via the shared `resolveModelConfig`
(`dispatch_model.go`):
- **`resolveExtractorChatConfig`** — UUID path calls
`resolveModelConfigByID` directly (one DB hit); composite path goes
through `resolveModelConfig`. Added `isBareTenantModelID` pre-check for
clear errors when a UUID doesn't exist.
- **`resolveExtractorChatTarget`** — propagates resolution errors
instead of silently returning empty driver.
- **`Chat()`** — removed `driver = "dummy"` fallback. Missing driver is
now an explicit error.
- **Removed dead code**: `splitExtractorLLID`,
`findExtractorSoleActiveInstance`.
### 4. `InjectExtractorLLMID` — fallback when no user config
(`internal/common/parser_config.go`)
Injects the tenant's global default LLM into extractor components **only
when their `llm_id` is empty**. Preserves user-selected UUID or
model@provider values.
Priority: user-configured llm_id > tenant global default > error (no
silent dummy fallback).
### 5. `ResponseHeaderTimeout` increase
(`internal/entity/models/base_model.go`)
`ResponseHeaderTimeout` 60s → 120s in `NewDriverHTTPClient`. Reasoning
models with large extraction prompts can take longer than 60s to produce
the first response token.
### 6. CJK rune-aware heading detection
(`internal/ingestion/component/chunker/title.go`)
Two byte-vs-rune bugs that only manifest on CJK text:
| Fix | What changed | Why |
|-----|-------------|-----|
| **`isColonTitle` byte offset** | `body[lastPunct+1:]` →
`body[lastPunct+runeLen:]` via `utf8.DecodeRuneInString` |
`strings.LastIndexAny` returns a byte index; `+1` skips only 1 byte,
corrupting multi-byte CJK punctuation (e.g. `。` = 3 bytes) and inflating
the rune count past the 32-rune threshold → false-positive heading
promotion. |
| **Short-line filter byte count** | `len(text) <= 1` →
`utf8.RuneCountInString(text) <= 1` | Go `len` is UTF-8 bytes; a single
CJK char (3 bytes) passed the filter, but Python's `len` returns 1 →
mismatch. |
### 7. `extractor_tag.go` — log error when llm fails
When `resolveExtractorChatTarget` returned an error, `runAutoTags` will
log error.
### 8. Python DSL prompt-key compatibility
(`internal/ingestion/component/extractor.go`)
The Resume DSL template uses Python-side key names (`sys_prompt`,
`prompts`). `NewExtractorComponent` now accepts them as fallbacks
alongside the Go names:
- `system_prompt` (Go) ← `sys_prompt` (Python) as fallback
- `prompt` (Go string) ← `prompts` (Python array `[{"role","content"}]`,
takes `[0].content`) as fallback
Mirrors the alias pattern already in `internal/agent/component/llm.go`.
`resolveInputs` accepts per-call `sys_prompt` override too.
## Remaining gaps vs Python
| Gap | Scope | Impact |
|-----|-------|--------|
| **TOC removal for TXT/MD/HTML** | Python's `remove_contents_table`
works on all text formats; Go's `remove_toc` is PDF-only. | Low —
plain-text documents rarely contain structured TOCs. |
| **Regex pattern details** | Minor differences in quantifiers, missing
H5/H6 markdown patterns, missing 4-level numbering pattern. | Low — Go's
variants are stricter; DOCX headings are covered by `ck_type` fallback.
|
## Testing
- `TestHierarchyTitleChunker_CKTypeHeadingFallback` — DOCX `ck_type`
heading promotion
- `TestHierarchyTitleChunker_ColonTitlePromotion` /
`_ColonTitleShortLine_Negative` — colon-title promotion + guard
- `TestIsColonTitle_CJKEdgeCase` / `TestIsColonTitle_ASCII_NoRegression`
— CJK byte-offset fix + ASCII regression
- `TestHierarchyTitleChunker_ColonTitlePromotion_CJK_EdgeCase` — CJK
colon edge case through full pipeline
- `TestHierarchyTitleChunker_ShortSingleCJKLineFilter` — single CJK char
filtered to body
- `TestHierarchyTitleChunker_ShortNumericLineFilter` — purely numeric
lines filtered
- `TestGetFileType_ImageExtensions` / `_ExistingFormats_NoRegression` —
image extension mapping
- `TestInjectExtractorLLMID_SkipWhenUUID` / `_SkipWhenComposite` /
`_InjectWhenEmpty` — llm_id injection guard
- `TestIsBareTenantModelID` — UUID detection
- `TestResolveExtractorChatTarget_AtSplitFallback` / `_NoDriver` — @
split fallback without DB
- `TestNewExtractorComponent_SysPromptAlias` / `_PromptsArray` /
`_PromptsArray_PromptWins` / `_SystemPromptWinsOverSysPrompt` — Python
key compatibility
- `TestBuildDOCXJSONSections_List` / `_TextBox` / `_MixedWithList` —
DOCX list/text_box parsing
- Full ingestion test suite passes (chunker, pipeline, task, service,
component packages)
2026-07-22 19:14:32 +08:00
|
|
|
Content: json.RawMessage(`[{"type":"text","text":"hello"}]`),
|
fix: docx/email parsing, extractor LLM driver, and chunker alignment (#17144)
## Summary
Three groups of changes across the Go ingestion pipeline:
### 1. DOCX parsing improvements
- **docx_parser.go**: Enhanced DOCX parsing with better structure
extraction and media handling
- **docx_parser_cgo_test.go**, **docx_parser_test.go**: Companion tests
- **office_parsers_no_cgo.go**: Stub sync for non-CGO builds
### 2. Email (.eml) parsing: base64 Content-Transfer-Encoding decoding
- **email_parser.go** (`decodeCTE`): Added Content-Transfer-Encoding
decoding for base64 and quoted-printable. Go's `mime/multipart.Reader`
does not decode Content-Transfer-Encoding automatically, so attachments
with `Content-Transfer-Encoding: base64` remained base64-encoded in the
output. The new `decodeCTE` helper is called after reading each
multipart part's raw bytes in `readMailBody`, mirroring Python's
`part.get_payload(decode=True)`.
- **email_parser_test.go**: Two new tests — simple base64 attachment and
nested multipart/alternative with base64 attachment.
### 3. Extractor LLM driver fix + ModelDriver consolidation
- **extractor.go**: Fixed a bug where the Extractor component used
`ModelFactory.CreateModelDriver()`, which creates bare model instances
without API keys or provider configuration. Switched to
`models.GetPreconfiguredDriver()` which resolves the actual
pre-configured driver from `ProviderManager`, matching the codepath used
by `llm.go`. This fixes auto keyword/question extraction in DSL
pipelines that require LLM calls.
- **get_driver.go** (new): Extracted shared `GetPreconfiguredDriver()`
from `llm.go:newChatModelDriver()` so both `llm.go` and `extractor.go`
use the same codepath.
- **get_driver_test.go** (new): Tests for the shared driver resolution.
- **llm.go**: Replaced inline driver resolution with
`models.GetPreconfiguredDriver()`.
### 4. Chunker fixes and observability
- **group.go** (`extractLineRecords`): Fixed to also read `markdown` and
`html` payload keys — previously it only read `text`/`content`, causing
GroupTitleChunker to silently return empty results for markdown-format
parser output.
- **common.go** (`compileDelimPattern`): Aligned with Python's
`_compile_delimiter_pattern` — only backtick-wrapped delimiters produce
an active regex pattern; plain delimiters are not compiled into the
split regex.
- **token.go** (`applyChildrenDelim`): Set `DocType` and `CKType` to
`"text"` on created ChunkDocs so the token-size merge path correctly
identifies and merges text segments.
- **parser.go**, **extractor.go**, **tokenizer.go**, **group.go**,
**hierarchy.go**: Added debug-level logging for pipeline diagnostics.
- **parser_dispatch_test.go**, **group_test.go**: New tests.
## Verification
- All Go tests pass: `bash build.sh --test ./internal/parser/parser/...`
and `bash build.sh --test ./internal/ingestion/component/...`
- Build succeeds: `bash build.sh --go`
2026-07-21 13:51:17 +08:00
|
|
|
}},
|
|
|
|
|
}},
|
|
|
|
|
}},
|
|
|
|
|
}
|
|
|
|
|
html := docxIRTableToHTML(el)
|
|
|
|
|
want := "<table><tr><td>hello</td></tr></table>"
|
|
|
|
|
if html != want {
|
|
|
|
|
t.Errorf("single cell html = %q, want %q", html, want)
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func TestDocxIRTableToHTML_MultiRowCol(t *testing.T) {
|
|
|
|
|
cell := func(text string) docxIRCell {
|
|
|
|
|
return docxIRCell{
|
|
|
|
|
Content: []docxIRElement{{
|
|
|
|
|
Type: "paragraph",
|
fix(ingestion): align laws DSL with Python — heading fallback, colon-title, short-line filter, remove_toc, and image extension mapping (#17200)
## Summary
This PR aligns the Go ingestion pipeline's **Laws** DSL template with
the Python implementation by fixing heading-detection gaps, adds
image-extension support, refactors the **Extractor** component's LLM
resolution, hardens heading detection for CJK text, and makes the
Extractor accept the Python DSL prompt key names
(`sys_prompt`/`prompts`) alongside the Go names.
## Changes
### 1. Picture file-type detection (`internal/utility/file.go`)
Adds explicit mapping for common image extensions (png, jpg, jpeg, gif,
bmp, tiff, tif, webp, svg, ico, avif, heic, apng) → `FileTypeVISUAL`,
with regression tests.
### 2. Laws DSL heading-detection alignment
(`internal/ingestion/component/chunker/`)
Four fixes to `resolveTitleLevels`:
| Fix | What changed | Why |
|-----|-------------|-----|
| **DOCX `ck_type` fallback** | `ckType` field on `lineRecord`,
propagated from `ChunkDoc.CKType` in `recordsFromStructured`. When
`ck_type=="heading"`, assign `fallbackLevel`. | office_oxide extracts
DOCX heading metadata, but the info was lost before reaching the heading
detector. Word headings whose text doesn't match any regex (e.g.
"Introduction") were treated as body. |
| **`make_colon_as_title` promotion** | `isColonTitle()`: promotes lines
ending with `:`/`:` that have sentence-ending punctuation before the
colon and ≥32 runes between them. | Mirrors Python's
`make_colon_as_title` in `rag/nlp/__init__.py`. Triple guard prevents
false positives. |
| **Short/numeric line filter** | Lines with ≤1 rune or purely numeric
are pinned to body level. | Mirrors Python `tree_merge`'s filter of
`sections` where `len(...) <= 1` or `re.match(r"[0-9]+$", ...)`. |
| **PDF `remove_toc`** | `"remove_toc": true` added to the PDF parser
setup in `ingestion_pipeline_laws.json`. | The Go PDF parser already
supports TOC removal; the Book template already enables it. |
### 3. Extractor llm_id resolution
(`internal/ingestion/component/extractor.go`)
Refactored to handle both **bare tenant_model UUIDs** and **composite
model@provider** strings via the shared `resolveModelConfig`
(`dispatch_model.go`):
- **`resolveExtractorChatConfig`** — UUID path calls
`resolveModelConfigByID` directly (one DB hit); composite path goes
through `resolveModelConfig`. Added `isBareTenantModelID` pre-check for
clear errors when a UUID doesn't exist.
- **`resolveExtractorChatTarget`** — propagates resolution errors
instead of silently returning empty driver.
- **`Chat()`** — removed `driver = "dummy"` fallback. Missing driver is
now an explicit error.
- **Removed dead code**: `splitExtractorLLID`,
`findExtractorSoleActiveInstance`.
### 4. `InjectExtractorLLMID` — fallback when no user config
(`internal/common/parser_config.go`)
Injects the tenant's global default LLM into extractor components **only
when their `llm_id` is empty**. Preserves user-selected UUID or
model@provider values.
Priority: user-configured llm_id > tenant global default > error (no
silent dummy fallback).
### 5. `ResponseHeaderTimeout` increase
(`internal/entity/models/base_model.go`)
`ResponseHeaderTimeout` 60s → 120s in `NewDriverHTTPClient`. Reasoning
models with large extraction prompts can take longer than 60s to produce
the first response token.
### 6. CJK rune-aware heading detection
(`internal/ingestion/component/chunker/title.go`)
Two byte-vs-rune bugs that only manifest on CJK text:
| Fix | What changed | Why |
|-----|-------------|-----|
| **`isColonTitle` byte offset** | `body[lastPunct+1:]` →
`body[lastPunct+runeLen:]` via `utf8.DecodeRuneInString` |
`strings.LastIndexAny` returns a byte index; `+1` skips only 1 byte,
corrupting multi-byte CJK punctuation (e.g. `。` = 3 bytes) and inflating
the rune count past the 32-rune threshold → false-positive heading
promotion. |
| **Short-line filter byte count** | `len(text) <= 1` →
`utf8.RuneCountInString(text) <= 1` | Go `len` is UTF-8 bytes; a single
CJK char (3 bytes) passed the filter, but Python's `len` returns 1 →
mismatch. |
### 7. `extractor_tag.go` — log error when llm fails
When `resolveExtractorChatTarget` returned an error, `runAutoTags` will
log error.
### 8. Python DSL prompt-key compatibility
(`internal/ingestion/component/extractor.go`)
The Resume DSL template uses Python-side key names (`sys_prompt`,
`prompts`). `NewExtractorComponent` now accepts them as fallbacks
alongside the Go names:
- `system_prompt` (Go) ← `sys_prompt` (Python) as fallback
- `prompt` (Go string) ← `prompts` (Python array `[{"role","content"}]`,
takes `[0].content`) as fallback
Mirrors the alias pattern already in `internal/agent/component/llm.go`.
`resolveInputs` accepts per-call `sys_prompt` override too.
## Remaining gaps vs Python
| Gap | Scope | Impact |
|-----|-------|--------|
| **TOC removal for TXT/MD/HTML** | Python's `remove_contents_table`
works on all text formats; Go's `remove_toc` is PDF-only. | Low —
plain-text documents rarely contain structured TOCs. |
| **Regex pattern details** | Minor differences in quantifiers, missing
H5/H6 markdown patterns, missing 4-level numbering pattern. | Low — Go's
variants are stricter; DOCX headings are covered by `ck_type` fallback.
|
## Testing
- `TestHierarchyTitleChunker_CKTypeHeadingFallback` — DOCX `ck_type`
heading promotion
- `TestHierarchyTitleChunker_ColonTitlePromotion` /
`_ColonTitleShortLine_Negative` — colon-title promotion + guard
- `TestIsColonTitle_CJKEdgeCase` / `TestIsColonTitle_ASCII_NoRegression`
— CJK byte-offset fix + ASCII regression
- `TestHierarchyTitleChunker_ColonTitlePromotion_CJK_EdgeCase` — CJK
colon edge case through full pipeline
- `TestHierarchyTitleChunker_ShortSingleCJKLineFilter` — single CJK char
filtered to body
- `TestHierarchyTitleChunker_ShortNumericLineFilter` — purely numeric
lines filtered
- `TestGetFileType_ImageExtensions` / `_ExistingFormats_NoRegression` —
image extension mapping
- `TestInjectExtractorLLMID_SkipWhenUUID` / `_SkipWhenComposite` /
`_InjectWhenEmpty` — llm_id injection guard
- `TestIsBareTenantModelID` — UUID detection
- `TestResolveExtractorChatTarget_AtSplitFallback` / `_NoDriver` — @
split fallback without DB
- `TestNewExtractorComponent_SysPromptAlias` / `_PromptsArray` /
`_PromptsArray_PromptWins` / `_SystemPromptWinsOverSysPrompt` — Python
key compatibility
- `TestBuildDOCXJSONSections_List` / `_TextBox` / `_MixedWithList` —
DOCX list/text_box parsing
- Full ingestion test suite passes (chunker, pipeline, task, service,
component packages)
2026-07-22 19:14:32 +08:00
|
|
|
Content: rawJSON([]docxIRRun{{Type: "text", Text: text}}),
|
fix: docx/email parsing, extractor LLM driver, and chunker alignment (#17144)
## Summary
Three groups of changes across the Go ingestion pipeline:
### 1. DOCX parsing improvements
- **docx_parser.go**: Enhanced DOCX parsing with better structure
extraction and media handling
- **docx_parser_cgo_test.go**, **docx_parser_test.go**: Companion tests
- **office_parsers_no_cgo.go**: Stub sync for non-CGO builds
### 2. Email (.eml) parsing: base64 Content-Transfer-Encoding decoding
- **email_parser.go** (`decodeCTE`): Added Content-Transfer-Encoding
decoding for base64 and quoted-printable. Go's `mime/multipart.Reader`
does not decode Content-Transfer-Encoding automatically, so attachments
with `Content-Transfer-Encoding: base64` remained base64-encoded in the
output. The new `decodeCTE` helper is called after reading each
multipart part's raw bytes in `readMailBody`, mirroring Python's
`part.get_payload(decode=True)`.
- **email_parser_test.go**: Two new tests — simple base64 attachment and
nested multipart/alternative with base64 attachment.
### 3. Extractor LLM driver fix + ModelDriver consolidation
- **extractor.go**: Fixed a bug where the Extractor component used
`ModelFactory.CreateModelDriver()`, which creates bare model instances
without API keys or provider configuration. Switched to
`models.GetPreconfiguredDriver()` which resolves the actual
pre-configured driver from `ProviderManager`, matching the codepath used
by `llm.go`. This fixes auto keyword/question extraction in DSL
pipelines that require LLM calls.
- **get_driver.go** (new): Extracted shared `GetPreconfiguredDriver()`
from `llm.go:newChatModelDriver()` so both `llm.go` and `extractor.go`
use the same codepath.
- **get_driver_test.go** (new): Tests for the shared driver resolution.
- **llm.go**: Replaced inline driver resolution with
`models.GetPreconfiguredDriver()`.
### 4. Chunker fixes and observability
- **group.go** (`extractLineRecords`): Fixed to also read `markdown` and
`html` payload keys — previously it only read `text`/`content`, causing
GroupTitleChunker to silently return empty results for markdown-format
parser output.
- **common.go** (`compileDelimPattern`): Aligned with Python's
`_compile_delimiter_pattern` — only backtick-wrapped delimiters produce
an active regex pattern; plain delimiters are not compiled into the
split regex.
- **token.go** (`applyChildrenDelim`): Set `DocType` and `CKType` to
`"text"` on created ChunkDocs so the token-size merge path correctly
identifies and merges text segments.
- **parser.go**, **extractor.go**, **tokenizer.go**, **group.go**,
**hierarchy.go**: Added debug-level logging for pipeline diagnostics.
- **parser_dispatch_test.go**, **group_test.go**: New tests.
## Verification
- All Go tests pass: `bash build.sh --test ./internal/parser/parser/...`
and `bash build.sh --test ./internal/ingestion/component/...`
- Build succeeds: `bash build.sh --go`
2026-07-21 13:51:17 +08:00
|
|
|
}},
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
el := docxIRElement{
|
|
|
|
|
Type: "table",
|
|
|
|
|
Rows: []docxIRRow{
|
|
|
|
|
{Cells: []docxIRCell{cell("A1"), cell("B1")}},
|
|
|
|
|
{Cells: []docxIRCell{cell("A2"), cell("B2")}},
|
|
|
|
|
},
|
|
|
|
|
}
|
|
|
|
|
html := docxIRTableToHTML(el)
|
|
|
|
|
want := "<table><tr><td>A1</td><td>B1</td></tr><tr><td>A2</td><td>B2</td></tr></table>"
|
|
|
|
|
if html != want {
|
|
|
|
|
t.Errorf("multi cell html = %q, want %q", html, want)
|
|
|
|
|
}
|
|
|
|
|
}
|