Nomad changes

fixtures/contract/api_batch_bytes_async.json

@@ -0,0 +1,30 @@
+{
+  "id": "api_batch_bytes_async",
+  "description": "Tests async batch bytes extraction API (batch_extract_bytes)",
+  "tags": ["contract", "api", "batch"],
+  "call": "extract_file",
+  "skip": {
+    "languages": ["wasm"],
+    "reason": "WASM uses synchronous-only API; async extraction is not available on the wasm target"
+  },
+  "input": {
+    "path": "pdf/fake_memo.pdf"
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "contains_any",
+      "field": "content",
+      "values": ["May 5, 2023", "Mallori"]
+    }
+  ]
+}

fixtures/contract/api_batch_bytes_with_configs_async.json

@@ -0,0 +1,33 @@
+{
+  "id": "api_batch_bytes_with_configs_async",
+  "description": "Tests async batch bytes extraction with per-file configs (batch_extract_bytes with file_configs parameter)",
+  "tags": ["contract", "api", "batch", "file_config"],
+  "call": "extract_file",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "output_format": "markdown"
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "equals",
+      "field": "metadata.output_format",
+      "value": "markdown"
+    }
+  ],
+  "skip": {
+    "languages": ["wasm"],
+    "reason": "WASM uses synchronous-only API; async extraction is not available on the wasm target"
+  }
+}

fixtures/contract/api_batch_file_async.json

@@ -0,0 +1,30 @@
+{
+  "id": "api_batch_file_async",
+  "description": "Tests async batch file extraction API (batch_extract_file)",
+  "tags": ["contract", "api", "batch"],
+  "call": "extract_file",
+  "input": {
+    "path": "pdf/fake_memo.pdf"
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "contains_any",
+      "field": "content",
+      "values": ["May 5, 2023", "Mallori"]
+    }
+  ],
+  "skip": {
+    "languages": ["wasm"],
+    "reason": "WASM uses synchronous-only API; async extraction is not available on the wasm target"
+  }
+}

fixtures/contract/api_batch_file_with_configs_async.json

@@ -0,0 +1,33 @@
+{
+  "id": "api_batch_file_with_configs_async",
+  "description": "Tests async batch file extraction with per-file configs (batch_extract_files with file_configs parameter)",
+  "tags": ["contract", "api", "batch", "file_config"],
+  "call": "extract_file",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "output_format": "markdown"
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "equals",
+      "field": "metadata.output_format",
+      "value": "markdown"
+    }
+  ],
+  "skip": {
+    "languages": ["wasm"],
+    "reason": "WASM uses synchronous-only API; async extraction is not available on the wasm target"
+  }
+}

fixtures/contract/api_extract_bytes_async.json

@@ -0,0 +1,30 @@
+{
+  "id": "api_extract_bytes_async",
+  "description": "Tests async bytes extraction API (extract_bytes)",
+  "tags": ["contract", "api"],
+  "call": "extract_file",
+  "input": {
+    "path": "pdf/fake_memo.pdf"
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "contains_any",
+      "field": "content",
+      "values": ["May 5, 2023", "Mallori"]
+    }
+  ],
+  "skip": {
+    "languages": ["wasm"],
+    "reason": "WASM uses synchronous-only API; async extraction is not available on the wasm target"
+  }
+}

fixtures/contract/api_extract_file_async.json

@@ -0,0 +1,29 @@
+{
+  "id": "api_extract_file_async",
+  "description": "Tests async file extraction API (extract_file)",
+  "tags": ["contract", "api"],
+  "input": {
+    "path": "pdf/fake_memo.pdf"
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "contains_any",
+      "field": "content",
+      "values": ["May 5, 2023", "Mallori"]
+    }
+  ],
+  "skip": {
+    "languages": ["wasm"],
+    "reason": "WASM uses synchronous-only API; async extraction is not available on the wasm target"
+  }
+}

fixtures/contract/config_chunking_prepend_heading_context.json

@@ -0,0 +1,52 @@
+{
+  "id": "config_chunking_prepend_heading_context",
+  "description": "Tests markdown chunker prepends heading hierarchy to chunk content",
+  "tags": [
+    "contract",
+    "config",
+    "chunking",
+    "heading-context"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "markdown/extraction_test.md",
+    "config": {
+      "chunking": {
+        "chunker_type": "markdown",
+        "max_chars": 300,
+        "max_overlap": 50,
+        "prepend_heading_context": true
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "count_min",
+      "field": "chunks",
+      "value": 2
+    },
+    {
+      "type": "is_true",
+      "field": "chunks_have_content"
+    },
+    {
+      "type": "is_true",
+      "field": "chunks_have_heading_context"
+    },
+    {
+      "type": "is_true",
+      "field": "first_chunk_starts_with_heading"
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_document_structure_with_headings.json

@@ -0,0 +1,37 @@
+{
+  "id": "config_document_structure_with_headings",
+  "description": "Tests document structure with DOCX heading-driven nesting",
+  "tags": [
+    "contract",
+    "document_structure"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "docx/fake.docx",
+    "config": {
+      "include_document_structure": true
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+    },
+    {
+      "type": "not_empty",
+      "field": "document"
+    },
+    {
+      "type": "count_min",
+      "field": "document.nodes",
+      "value": 1
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_element_types.json

@@ -0,0 +1,36 @@
+{
+  "id": "config_element_types",
+  "description": "Tests element-based result format with element type assertions on DOCX",
+  "tags": [
+    "contract",
+    "config",
+    "result_format"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "docx/unit_test_headers.docx",
+    "config": {
+      "result_format": "element_based"
+    }
+  },
+  "assertions": [
+    {
+      "type": "contains_any",
+      "field": "mime_type",
+      "values": [
+        "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+      ]
+    },
+    {
+      "type": "count_min",
+      "field": "elements",
+      "value": 1
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_embedding_plugin.json

@@ -0,0 +1,32 @@
+{
+  "id": "config_embedding_plugin",
+  "description": "Tests EmbeddingModelType::Plugin variant deserialization in ChunkingConfig — config accepts the plugin variant shape; actual dispatch requires a host-language backend registered via register_embedding_backend at runtime",
+  "tags": ["contract", "config", "embeddings", "plugin"],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "chunking": {
+        "max_chars": 500,
+        "max_overlap": 50,
+        "embedding": {
+          "model": {
+            "type": "plugin",
+            "name": "test-plugin-backend"
+          },
+          "normalize": true,
+          "max_embed_duration_secs": 30
+        }
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    }
+  ],
+  "skip": {
+    "languages": ["python", "rust", "node", "go", "ruby", "elixir", "wasm", "java", "csharp", "php", "r", "dart", "kotlin_android", "swift", "zig"],
+    "reason": "EmbeddingModelType::Plugin requires a host-language backend registered via register_embedding_backend before dispatch; the e2e harness cannot register one. This fixture validates config round-trip (the {\"type\":\"plugin\",\"name\":...} shape is accepted by every binding's EmbeddingConfig)."
+  }
+}

fixtures/contract/config_extraction_timeout.json

@@ -0,0 +1,34 @@
+{
+  "id": "config_extraction_timeout",
+  "description": "Tests that extraction_timeout_secs config field is accepted and does not affect fast extractions",
+  "tags": [
+    "contract",
+    "config",
+    "timeout"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "extraction_timeout_secs": 300
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_keywords.json

@@ -0,0 +1,46 @@
+{
+  "id": "config_keywords",
+  "description": "Tests keyword extraction via YAKE algorithm",
+  "tags": [
+    "contract",
+    "config",
+    "keywords"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "keywords": {
+        "algorithm": "yake",
+        "max_keywords": 10
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "not_empty",
+      "field": "keywords"
+    },
+    {
+      "type": "count_min",
+      "field": "keywords",
+      "value": 1
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_llm_structured_extraction.json

@@ -0,0 +1,52 @@
+{
+  "id": "config_llm_structured_extraction",
+  "description": "Tests structured extraction via liter-llm with JSON schema",
+  "tags": ["contract", "config", "liter-llm", "structured-extraction"],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "structured_extraction": {
+        "schema": {
+          "type": "object",
+          "properties": {
+            "title": {
+              "type": "string"
+            },
+            "date": {
+              "type": "string"
+            },
+            "summary": {
+              "type": "string"
+            }
+          },
+          "required": ["title"]
+        },
+        "schema_name": "memo_data",
+        "llm": {
+          "model": "openai/gpt-4o"
+        }
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "not_empty",
+      "field": "structured_output"
+    }
+  ],
+  "skip": {
+    "languages": ["python", "rust", "node", "go", "ruby", "elixir", "wasm", "java", "csharp", "php", "r", "dart", "kotlin_android", "swift", "zig"],
+    "reason": "Requires liter-llm feature and KREUZBERG_LLM_API_KEY env var; runtime-only skip"
+  }
+}

fixtures/contract/config_pages.json

@@ -0,0 +1,43 @@
+{
+  "id": "config_pages",
+  "description": "Tests page extraction and page marker configuration",
+  "tags": [
+    "contract",
+    "config"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "pages": {
+        "extract_pages": true,
+        "insert_page_markers": true
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "contains_any",
+      "field": "content",
+      "values": [
+        "PAGE"
+      ]
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_quality_enabled.json

@@ -0,0 +1,48 @@
+{
+  "id": "config_quality_enabled",
+  "description": "Tests quality scoring produces a score value in [0.0, 1.0]",
+  "tags": [
+    "contract",
+    "config",
+    "quality"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "enable_quality_processing": true
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "not_empty",
+      "field": "quality_score"
+    },
+    {
+      "type": "greater_than_or_equal",
+      "field": "quality_score",
+      "value": 0.0
+    },
+    {
+      "type": "less_than_or_equal",
+      "field": "quality_score",
+      "value": 1.0
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_security_limits.json

@@ -0,0 +1,41 @@
+{
+  "id": "config_security_limits",
+  "description": "Tests archive extraction with custom security limits",
+  "tags": [
+    "contract",
+    "config",
+    "security"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "archives/documents.zip",
+    "config": {
+      "security_limits": {
+        "max_archive_size": 104857600,
+        "max_compression_ratio": 50,
+        "max_files_in_archive": 100
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "contains_any",
+      "field": "mime_type",
+      "values": [
+        "application/zip",
+        "application/x-zip-compressed"
+      ]
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/config_tree_sitter.json

@@ -0,0 +1,51 @@
+{
+  "id": "config_tree_sitter",
+  "description": "Tests tree-sitter configuration round-trip",
+  "tags": [
+    "contract",
+    "config",
+    "tree-sitter"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "code/hello.py",
+    "config": {
+      "tree_sitter": {
+        "languages": [
+          "python",
+          "rust"
+        ],
+        "groups": [
+          "web"
+        ],
+        "process": {
+          "structure": true,
+          "imports": true,
+          "exports": true,
+          "comments": false,
+          "docstrings": false,
+          "symbols": false,
+          "diagnostics": false
+        }
+      }
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "text/x-source-code"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 5
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/contract/output_format_bytes_markdown.json

@@ -0,0 +1,30 @@
+{
+  "id": "output_format_bytes_markdown",
+  "description": "Tests markdown output format via bytes extraction API",
+  "tags": ["contract", "output_format", "bytes"],
+  "call": "extract_bytes_sync",
+  "input": {
+    "data": "pdf/fake_memo.pdf",
+    "mime_type": "application/pdf",
+    "config": {
+      "output_format": "markdown"
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "equals",
+      "field": "metadata.output_format",
+      "value": "markdown"
+    }
+  ]
+}

fixtures/contract/output_format_markdown.json

@@ -0,0 +1,38 @@
+{
+  "id": "output_format_markdown",
+  "description": "Tests Markdown output format",
+  "tags": [
+    "contract",
+    "output_format"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pdf/fake_memo.pdf",
+    "config": {
+      "output_format": "markdown"
+    }
+  },
+  "assertions": [
+    {
+      "type": "equals",
+      "field": "mime_type",
+      "value": "application/pdf"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 10
+    },
+    {
+      "type": "equals",
+      "field": "metadata.output_format",
+      "value": "markdown"
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}