Nomad changes

fixtures/format_specific/format_docx_standalone.json

@@ -0,0 +1,21 @@
+{
+  "id": "format_docx_standalone",
+  "category": "format_specific",
+  "description": "Standalone DOCX extraction using extract_bytes_sync",
+  "tags": ["format_specific", "docx", "text_extraction"],
+  "call": "extract_bytes_sync",
+  "input": {
+    "data": "docx/fake.docx",
+    "mime_type": "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 20
+    }
+  ]
+}

fixtures/format_specific/format_hwpx_standalone.json

@@ -0,0 +1,26 @@
+{
+  "id": "format_hwpx_standalone",
+  "category": "format_specific",
+  "description": "Standalone HWPX extraction using extract_bytes_sync",
+  "tags": ["format_specific", "hwpx", "text_extraction"],
+  "call": "extract_bytes_sync",
+  "input": {
+    "data": "hwpx/simple.hwpx",
+    "mime_type": "application/haansofthwpx"
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 20
+    },
+    {
+      "type": "contains",
+      "field": "content",
+      "value": "Hello from HWPX"
+    }
+  ]
+}

fixtures/format_specific/format_pdf_text.json

@@ -0,0 +1,26 @@
+{
+  "id": "format_pdf_text",
+  "category": "format_specific",
+  "description": "Standalone PDF text extraction using extract_bytes_sync",
+  "tags": ["format_specific", "pdf", "text_extraction"],
+  "call": "extract_bytes_sync",
+  "input": {
+    "data": "pdf/fake_memo.pdf",
+    "mime_type": "application/pdf"
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 50
+    },
+    {
+      "type": "contains_any",
+      "field": "content",
+      "values": ["Mallori", "May"]
+    }
+  ]
+}

fixtures/format_specific/format_pptx.json

@@ -0,0 +1,26 @@
+{
+  "id": "format_pptx",
+  "category": "format_specific",
+  "description": "PPTX presentation extraction using extract_file_sync",
+  "tags": [
+    "format_specific",
+    "pptx",
+    "text_extraction"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "pptx/simple.pptx",
+    "mime_type": "application/vnd.openxmlformats-officedocument.presentationml.presentation"
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}

fixtures/format_specific/format_xlsx.json

@@ -0,0 +1,26 @@
+{
+  "id": "format_xlsx",
+  "category": "format_specific",
+  "description": "XLSX spreadsheet extraction using extract_file_sync",
+  "tags": [
+    "format_specific",
+    "xlsx",
+    "text_extraction"
+  ],
+  "call": "extract_file_sync",
+  "input": {
+    "path": "xlsx/stanley_cups.xlsx",
+    "mime_type": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    }
+  ],
+  "skip": {
+    "languages": [
+      "wasm"
+    ],
+    "reason": "WASM cannot access filesystem; use extractBytes with file content instead"
+  }
+}