diff --git a/doc-converter/skills/doc-converter-expert/SKILL.md b/doc-converter/skills/doc-converter-expert/SKILL.md index 9608e91..834eb2d 100644 --- a/doc-converter/skills/doc-converter-expert/SKILL.md +++ b/doc-converter/skills/doc-converter-expert/SKILL.md @@ -75,6 +75,18 @@ FALSCH: Logo mehrfach im HTML anzeigen - Der Content beginnt DIREKT mit dem Dokumenttitel (`

`) und dann Sektion 1 - **VERBOTEN:** Klickbare Links wie "1. Zweck", "2. Geltungsbereich" im Content! +### Pipeline-Architektur (3-Phasen-Hybrid) + +``` +PDF → pdftoppm → Seitenbilder + Phase 1: recognize_document_structure() (Gemini, 3 Content-Seiten) + → header_bottom_pct, footer_top_pct (Crop-Koordinaten) + crop_page_images() → gecroppte Seitenbilder (Header/Footer entfernt) + Phase 2a: TOC + Footer generieren + Phase 2b: Content pro Seite (gecroppte Bilder, KEINE Header/Footer-Referenz) + Post-Processing: normalize_content_structure() → normalize_font_sizes() +``` + ### TOC-Sektion-1-Fix Wenn "1. Inhaltsverzeichnis" die erste Sektion im TOC ist: - Die Funktion `fix_toc_section_one()` korrigiert automatisch @@ -621,6 +633,35 @@ Automatische Normalisierung nach der Gemini-Konvertierung: --- +## Seitenbild-Cropping (Header/Footer-Entfernung) + +### Gemini-basierte Crop-Erkennung (Phase 1) +Phase 1 sendet Pre-Content + **3 Content-Seiten** an Gemini. Der RECOGNITION_PROMPT instruiert: +- "Vergleiche die Seiten. Alles was sich auf mind. 2 Seiten am oberen/unteren Rand wiederholt = Header/Footer" +- Seitenzahlen zählen auch dazu (Position gleich, nur Zahl ändert sich) +- `header_bottom_percent` und `footer_top_percent` sind **physische Crop-Grenzen** +- Beispiel: Footer-Trennlinie bei 88%, Text 89-91%, Balken 91-92% → `footer_top_percent = 87.5` + +### crop_page_images() — Physisches Cropping +Nach Phase 1 werden ALLE Seitenbilder physisch gecroppt: +```python +crop_page_images(page_images, header_bottom_pct, footer_top_pct, margin_px=5) +``` +- Header/Footer werden abgeschnitten bevor Gemini die Bilder in Phase 2b sieht +- Gemini kann NICHT durch Header/Footer verwirrt werden +- Prompts enthalten KEINE "IGNORIEREN!"-Sektionen mehr +- Stattdessen: "Die Seitenbilder zeigen NUR den Content-Bereich. Konvertiere ALLES." +- Original-Bilder bleiben für visuelle Verifikation erhalten + +### Vorteile des Croppings +1. **Keine Header/Footer-Artefakte möglich** (Gemini sieht sie nie) +2. **Kürzere Prompts** (keine "KOPFZEILE IGNORIEREN" Sektionen) +3. **Keine Header/Footer-Referenzbilder** → weniger Tokens → günstiger +4. **Tabellenfortsetzung einfacher** (Content reicht bis unterer Rand = Tabelle geht weiter) +5. **Weniger Post-Processing** nötig + +--- + ## Bild-Verarbeitung ### Waifu2x Upscaling diff --git a/n8n-skills/skills/code-javascript/SKILL.md b/n8n-skills/skills/code-javascript/SKILL.md index 7d7e8da..449a7b6 100644 --- a/n8n-skills/skills/code-javascript/SKILL.md +++ b/n8n-skills/skills/code-javascript/SKILL.md @@ -1,5 +1,5 @@ --- -name: code-javascript +name: n8n-code-javascript description: Write JavaScript code in n8n Code nodes. Use when writing JavaScript in n8n, using $input/$json/$node syntax, making HTTP requests with $helpers, working with dates using DateTime, troubleshooting Code node errors, or choosing between Code node modes. --- diff --git a/n8n-skills/skills/code-python/SKILL.md b/n8n-skills/skills/code-python/SKILL.md index 87568d2..67e8816 100644 --- a/n8n-skills/skills/code-python/SKILL.md +++ b/n8n-skills/skills/code-python/SKILL.md @@ -1,5 +1,5 @@ --- -name: code-python +name: n8n-code-python description: Write Python code in n8n Code nodes. Use when writing Python in n8n, using _input/_json/_node syntax, working with standard library, or need to understand Python limitations in n8n Code nodes. --- diff --git a/n8n-skills/skills/expression-syntax/SKILL.md b/n8n-skills/skills/expression-syntax/SKILL.md index b5eda81..2276cdb 100644 --- a/n8n-skills/skills/expression-syntax/SKILL.md +++ b/n8n-skills/skills/expression-syntax/SKILL.md @@ -1,5 +1,5 @@ --- -name: expression-syntax +name: n8n-expression-syntax description: Validate n8n expression syntax and fix common errors. Use when writing n8n expressions, using {{}} syntax, accessing $json/$node variables, troubleshooting expression errors, or working with webhook data in workflows. --- diff --git a/n8n-skills/skills/mcp-tools-expert/SKILL.md b/n8n-skills/skills/mcp-tools-expert/SKILL.md index 3e03989..090a877 100644 --- a/n8n-skills/skills/mcp-tools-expert/SKILL.md +++ b/n8n-skills/skills/mcp-tools-expert/SKILL.md @@ -1,5 +1,5 @@ --- -name: mcp-tools-expert +name: n8n-mcp-tools-expert description: Expert guide for using n8n-mcp MCP tools effectively. Use when searching for nodes, validating configurations, accessing templates, managing workflows, or using any n8n-mcp tool. Provides tool selection guidance, parameter formats, and common patterns. --- diff --git a/n8n-skills/skills/node-configuration/SKILL.md b/n8n-skills/skills/node-configuration/SKILL.md index ac0383c..8bc4753 100644 --- a/n8n-skills/skills/node-configuration/SKILL.md +++ b/n8n-skills/skills/node-configuration/SKILL.md @@ -1,5 +1,5 @@ --- -name: node-configuration +name: n8n-node-configuration description: Operation-aware node configuration guidance. Use when configuring nodes, understanding property dependencies, determining required fields, choosing between get_node detail levels, or learning common configuration patterns by node type. --- diff --git a/n8n-skills/skills/validation-expert/SKILL.md b/n8n-skills/skills/validation-expert/SKILL.md index a0249b2..fd6ce22 100644 --- a/n8n-skills/skills/validation-expert/SKILL.md +++ b/n8n-skills/skills/validation-expert/SKILL.md @@ -1,5 +1,5 @@ --- -name: validation-expert +name: n8n-validation-expert description: Interpret validation errors and guide fixing them. Use when encountering validation errors, validation warnings, false positives, operator structure issues, or need help understanding validation results. Also use when asking about validation profiles, error types, or the validation loop process. --- diff --git a/n8n-skills/skills/workflow-patterns/SKILL.md b/n8n-skills/skills/workflow-patterns/SKILL.md index 850151a..437bc5d 100644 --- a/n8n-skills/skills/workflow-patterns/SKILL.md +++ b/n8n-skills/skills/workflow-patterns/SKILL.md @@ -1,5 +1,5 @@ --- -name: workflow-patterns +name: n8n-workflow-patterns description: Proven workflow architectural patterns from real n8n workflows. Use when building new workflows, designing workflow structure, choosing workflow patterns, planning workflow architecture, or asking about webhook processing, HTTP API integration, database operations, AI agent workflows, or scheduled tasks. ---