footprinter-cli 1.0.0rc1__py3-none-any.whl

footprinter/bundled/patterns/extensions.yaml

@@ -0,0 +1,283 @@
+# Extension-to-ContentType mappings for v2 classification
+# Each key maps to a list of file extensions (with leading dots)
+
+code:
+  # Standard programming languages
+  - '.py'
+  - '.js'
+  - '.ts'
+  - '.jsx'
+  - '.tsx'
+  - '.java'
+  - '.cls'
+  - '.trigger'
+  - '.apex'
+  - '.go'
+  - '.rs'
+  - '.rb'
+  - '.sh'
+  - '.bash'
+  - '.sql'
+  - '.c'
+  - '.cpp'
+  - '.h'
+  - '.hpp'
+  - '.swift'
+  - '.kt'
+  - '.scala'
+  - '.r'
+  - '.pl'
+  - '.php'
+  - '.lua'
+  - '.vim'
+  - '.el'
+  - '.zsh'
+  - '.ps1'
+  - '.psm1'
+  - '.css'
+  - '.scss'
+  - '.less'
+  - '.vue'
+  - '.svelte'
+  - '.ex'
+  - '.exs'
+  - '.erl'
+  - '.hs'
+  # JavaScript/TypeScript module variants
+  - '.mjs'
+  - '.cjs'
+  - '.mts'
+  - '.cts'
+  # Python type stubs and Cython
+  - '.pyi'
+  - '.pyx'
+  - '.pxd'
+  - '.pxi'
+  # Other languages common in packages
+  - '.f90'
+  - '.f'
+  - '.f95'
+  - '.for'
+  - '.cc'
+  - '.cxx'
+  - '.proto'
+  - '.typed'
+  - '.coffee'
+  - '.mdx'
+  - '.robot'
+  - '.wasm'
+  - '.doctest'
+  - '.fish'
+  - '.csh'
+  - '.wsdl'
+  # Build/project files
+  - '.dtd'
+  - '.sdl'
+  - '.pbxproj'
+  - '.xcscheme'
+  # Design files
+  - '.design'
+  # Test files
+  - '.test'
+  - '.spec'
+  # Init files
+  - '.init'
+
+image:
+  - '.png'
+  - '.jpg'
+  - '.jpeg'
+  - '.gif'
+  - '.webp'
+  - '.svg'
+  - '.ico'
+  - '.heic'
+  - '.heif'
+  - '.bmp'
+  - '.tiff'
+  - '.tif'
+  - '.avif'
+  - '.raw'
+  - '.cr2'
+  - '.nef'
+  - '.dng'
+  - '.psd'
+  - '.ai'
+  - '.eps'
+
+recording:
+  - '.mp4'
+  - '.mov'
+  - '.m4a'
+  - '.wav'
+  - '.mp3'
+  - '.webm'
+  - '.mkv'
+  - '.avi'
+  - '.m4v'
+  - '.flac'
+  - '.ogg'
+  - '.wmv'
+  - '.aac'
+  - '.wma'
+  - '.aiff'
+  - '.opus'
+  - '.zoom'
+
+transient:
+  - '.pyc'
+  - '.pyo'
+  - '.class'
+  - '.o'
+  - '.obj'
+  - '.exe'
+  - '.dll'
+  - '.so'
+  - '.dylib'
+  - '.log'
+  - '.tmp'
+  - '.temp'
+  - '.cache'
+  - '.lock'
+  - '.bak'
+  - '.swp'
+  - '.swo'
+  - '.map'
+  - '.tsbuildinfo'
+  - '.pysave'
+
+data:
+  - '.csv'
+  - '.xlsx'
+  - '.xls'
+  - '.parquet'
+  - '.tsv'
+  - '.sqlite'
+  - '.db'
+  - '.sql'
+  - '.mdb'
+  - '.accdb'
+  - '.dat'
+  - '.jsonl'
+  - '.ndjson'
+  # Font files (treated as data assets)
+  - '.woff'
+  - '.woff2'
+  - '.ttf'
+  - '.otf'
+  - '.eot'
+  # GIS shapefile components
+  - '.shp'
+  - '.shx'
+  - '.dbf'
+  - '.prj'
+  - '.sbn'
+  - '.sbx'
+  # Serialization formats
+  - '.pickle'
+  - '.pkl'
+  - '.jar'
+  - '.npy'
+  - '.npz'
+  # Playlist/media metadata
+  - '.m3u'
+  - '.m3u8'
+  - '.pls'
+  # Database/backup files
+  - '.sqlite3'
+  - '.bin'
+  # Package/installer files
+  - '.pkg'
+  - '.dmg'
+  - '.msi'
+  # Certificate files
+  - '.p12'
+  - '.crt'
+  - '.pem'
+  - '.cer'
+  - '.pfx'
+  - '.pkpass'
+  # Spreadsheet formats
+  - '.ods'
+  # Event/log files
+  - '.evt'
+  - '.evtx'
+  # Subtitle files
+  - '.srt'
+  - '.sub'
+  - '.vtt'
+  # Genealogy data
+  - '.ged'
+
+config:
+  - '.yaml'
+  - '.yml'
+  - '.toml'
+  - '.ini'
+  - '.conf'
+  - '.cfg'
+  - '.env'
+  - '.properties'
+  # Environment/config variants
+  - '.example'
+  - '.sample'
+  - '.template'
+  - '.local'
+  # IDE/editor workspace configs
+  - '.code-workspace'
+  - '.xcworkspacedata'
+  # Apple property lists
+  - '.plist'
+  # macOS automation
+  - '.agent'
+
+documentation:
+  - '.md'
+  - '.markdown'
+  - '.rst'
+  - '.adoc'
+  - '.asciidoc'
+  - '.tex'
+  - '.rtf'
+
+administrative:
+  - '.docx'
+  - '.doc'
+  - '.pdf'
+  - '.odt'
+  - '.pages'
+  - '.pptx'
+  - '.ppt'
+  - '.odp'
+  - '.key'
+  - '.numbers'
+  - '.ics'
+  - '.eml'
+  - '.email'
+  - '.msg'
+
+web:
+  - '.html'
+  - '.htm'
+  - '.xhtml'
+  - '.xml'
+  - '.xsl'
+  - '.xslt'
+  - '.svg'
+  - '.rss'
+  - '.atom'
+
+archive:
+  - '.zip'
+  - '.tar'
+  - '.gz'
+  - '.bz2'
+  - '.7z'
+  - '.rar'
+  - '.tgz'
+  - '.xz'
+  - '.dmg'
+  - '.iso'
+
+ambiguous:
+  - '.json'

footprinter/bundled/patterns/filename_patterns.yaml

@@ -0,0 +1,61 @@
+# Filename and path patterns for content type detection
+
+# TRANSIENT file patterns (safe to delete)
+transient_patterns:
+  - '\.DS_Store$'
+  - 'Thumbs\.db$'
+  - '\.tmp$'
+  - '\.temp$'
+  - '~$'
+  - '\.swp$'
+  - '\.swo$'
+  - '\.log$'
+  - '\.cache'
+  - '__pycache__'
+  - 'node_modules'
+  - '\.npm'
+  - '\.yarn'
+  - 'venv'
+  - '\.venv'
+  - 'env/'
+  - '\.env/'
+  - '\.pyc$'
+  - '\.pyo$'
+  - '\.class$'
+  - '\.o$'
+  - '\.obj$'
+  - '\.exe$'
+  - 'dist/'
+  - 'build/'
+  - '\.next/'
+  - '\.git/objects'
+  - '\.git/lfs'
+
+transient_path_patterns:
+  - '/Downloads/'
+  - '/Trash/'
+  - '/tmp/'
+  - '/temp/'
+  - '/cache/'
+  - '/\.cache/'
+  - '/Caches/'
+
+# REFERENCE file patterns (third-party docs, tutorials)
+reference_patterns:
+  - 'tutorial'
+  - 'manual'
+  - 'handbook'
+  - 'reference'
+  - 'documentation'
+  - 'getting[_-]?started'
+  - 'cheat[_-]?sheet'
+  - 'quick[_-]?start'
+  - 'user[_-]?guide'
+  - 'api[_-]?docs?'
+
+reference_path_patterns:
+  - '/tutorials?/'
+  - '/docs?/'
+  - '/reference/'
+  - '/manuals?/'
+  - '/guides?/'

footprinter/bundled/patterns/mime_mappings.yaml

@@ -0,0 +1,68 @@
+# MIME type fragment -> ContentType mapping
+# These appear in the database as partial MIME types
+# Values must match ContentType enum names
+
+# Audio/Video MIME fragments
+x-wav: MEDIA_RECORDING
+x-aiff: MEDIA_RECORDING
+mp4a-lat: MEDIA_RECORDING
+mpeg: MEDIA_RECORDING
+mp2t: MEDIA_RECORDING
+quicktim: MEDIA_RECORDING
+x-m4a: MEDIA_RECORDING
+audio: MEDIA_RECORDING
+video: MEDIA_RECORDING
+
+# Image MIME fragments
+x-icon: MEDIA_IMAGE
+x-png: MEDIA_IMAGE
+image: MEDIA_IMAGE
+
+# Code/text MIME fragments
+x-python: CODE
+javascri: CODE
+x-sh: CODE
+x-csh: CODE
+x-csrc: CODE
+x-c++sr: CODE
+x-java: CODE
+x-sql: CODE
+x-tex: DOCUMENTATION
+x-latex: DOCUMENTATION
+html: CODE
+xml: CODE
+plain: DOCUMENTATION
+rtf: DOCUMENTATION
+markdown: DOCUMENTATION
+calendar: ADMINISTRATIVE
+
+# Data MIME fragments
+octet-st: DATA
+spreadsh: DATA
+csv: DATA
+tab-sepa: DATA
+
+# Archive
+zip: DATA
+x-tar: DATA
+x-gzip: DATA
+
+# Document MIME fragments
+pdf: ADMINISTRATIVE
+msword: ADMINISTRATIVE
+wordproc: ADMINISTRATIVE
+presenta: ADMINISTRATIVE
+document: ADMINISTRATIVE
+vnd.ms-e: DATA
+vnd.oasi: ADMINISTRATIVE
+x-iwork-: ADMINISTRATIVE
+vnd.yell: CODE
+
+# Package/archive MIME types
+x-zip: DATA
+x-pkcs12: DATA
+x-x509-c: DATA
+x-shells: CODE
+wsdl+xml: CODE
+xml-dtd: CODE
+x-subrip: DATA

footprinter/bundled/patterns/salesforce_rules.yaml

@@ -0,0 +1,84 @@
+# Salesforce-specific classification rules
+# This file is a self-contained pattern pack. When enabled via pattern_packs
+# config, these extensions/patterns are merged into the classifier at init time.
+
+# Extensions representing portfolio work (KEEP)
+keep_extensions:
+  - '.cls'
+  - '.trigger'
+  - '.cmp'
+  - '.component'
+  - '.page'
+  - '.flow'
+  - '.workflow'
+  - '.object'
+  - '.layout'
+  - '.permissionset'
+  - '.profile'
+  - '.app'
+  - '.tab'
+
+# Salesforce-specific file extensions merged into core extension sets when pack is enabled
+extensions:
+  code:
+    - '.cmp'
+    - '.component'
+    - '.page'
+    - '.flow'
+    - '.workflow'
+    - '.object'
+    - '.layout'
+    - '.permissionset'
+    - '.profile'
+    - '.app'
+    - '.tab'
+    - '.quickaction'
+    - '.soql'
+    - '.tag'
+    - '.flexipage'
+    - '.report'
+    - '.objecttranslation'
+    - '.flowdefinition'
+    - '.globalvalueset'
+    - '.auradoc'
+    - '.topicsforobjects'
+  data:
+    - '.resource'
+
+# LWC/Aura paths (keep .js, .html, .css within these)
+component_paths:
+  - 'lwc/'
+  - 'aura/'
+  - 'force-app/'
+
+# Path markers that identify Salesforce project directories
+path_markers:
+  - 'force-app'
+  - '.sfdx'
+
+# LWC webruntime bundle file prefixes (no-extension files)
+lwr_file_prefixes:
+  - 'lwr_loader'
+  - 'lwr_app'
+  - 'lwr_error_shim'
+  - 'meta-pixel-event-dispatcher'
+  - 'runtime'
+
+# XML file path prefixes for Salesforce XML disambiguation
+xml_paths:
+  - 'force-app/'
+  - 'salesforce/'
+
+# Patterns for boilerplate/auto-generated files to EXCLUDE
+exclude_patterns:
+  - '-meta\.xml$'
+  - '\.settings$'
+  - '\.labels$'
+  - '\.sharingrules$'
+  - '\.assignmentrules$'
+  - '\.autoresponserules$'
+  - '\.escalationrules$'
+  - '\.approvalprocess$'
+  - '\.remotesite$'
+  - '\.namedcredential$'
+  - '\.externalcredential$'

footprinter/bundled/patterns/security_patterns.yaml

@@ -0,0 +1,27 @@
+# PII and credential detection patterns
+# Used by RetentionClassifier for sensitivity analysis
+
+pii_patterns:
+  # Identity numbers
+  ssn: '\b\d{3}[-\s]?\d{2}[-\s]?\d{4}\b'
+  ein: '\b\d{2}-\d{7}\b'
+  passport: '\b[A-Z]{1,2}\d{6,9}\b'
+  # Financial
+  credit_card: '\b(?:4[0-9]{12}(?:[0-9]{3})?|5[1-5][0-9]{14}|3[47][0-9]{13}|6(?:011|5[0-9]{2})[0-9]{12})\b'
+  bank_account: '\b(?:account|acct|routing)[\s#:]*\d{8,17}\b'
+  # Contact (contextual - only flag bulk lists, not single emails)
+  bulk_email_list: '(?:[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\s*[,;\n]){3,}'
+  phone: '\b(?:\+1[-.\s]?)?\(?\d{3}\)?[-.\s]?\d{3}[-.\s]?\d{4}\b'
+  # Address (full format only - street + city + state + zip)
+  address: '\b\d+\s+[\w\s]+(?:St(?:reet)?|Ave(?:nue)?|Rd|Road|Blvd|Dr(?:ive)?|Ln|Lane|Way|Ct|Court)[\s,]+[\w\s]+,?\s+[A-Z]{2}\s+\d{5}(?:-\d{4})?\b'
+  # Dates with context only
+  date_of_birth: '\b(?:DOB|Date of Birth|Birth(?:date)?|Born)[\s:]+\d{1,2}[/-]\d{1,2}[/-]\d{2,4}\b'
+
+credential_patterns:
+  api_key: "(?:api[_-]?key|apikey)\\s*[=:]\\s*[\"']?[\\w-]{20,}"
+  aws_key: 'AKIA[0-9A-Z]{16}'
+  aws_secret: "(?:aws[_-]?secret|secret[_-]?key)\\s*[=:]\\s*[\"']?[\\w/+=]{40}"
+  stripe_key: 'sk_(?:live|test)_[0-9a-zA-Z]{24,}'
+  password: "(?:password|passwd|pwd)\\s*[=:]\\s*[\"'][^\"']{6,}[\"']"
+  bearer_token: 'Bearer\s+[A-Za-z0-9_-]{20,}'
+  private_key: '-----BEGIN (?:RSA |EC )?PRIVATE KEY-----'

footprinter/bundled/samples/hidden-client-file-sample.txt

@@ -0,0 +1,2 @@
+Confidential memo from Hidden Client Ltd.
+This file is hidden via client-level cascade — should never appear in results.

footprinter/bundled/samples/opaque-project-file-sample.txt

@@ -0,0 +1,2 @@
+Notes from Opaque Project review.
+This file inherits opaque visibility from its parent project policy.

footprinter/bundled/samples/visible-file-sample.txt

@@ -0,0 +1,2 @@
+Sample document for Visible Project.
+This file is visible and readable — no restrictive policies apply.

footprinter/cli/__init__.py

@@ -0,0 +1,135 @@
+"""Footprinter CLI router."""
+
+import argparse
+import sys
+
+
+def _check_python_version() -> None:
+    """Exit with a clear message if Python is too old."""
+    if sys.version_info < (3, 11):
+        print(
+            f"Error: Footprinter requires Python 3.11 or later (found {sys.version_info[0]}.{sys.version_info[1]}).",
+            file=sys.stderr,
+        )
+        sys.exit(1)
+
+
+def _is_first_run() -> bool:
+    """Return True if neither config file nor database exists."""
+    from footprinter.paths import get_config_path, get_db_path
+
+    return not get_config_path().exists() and not get_db_path().exists()
+
+
+def main(argv=None) -> None:
+    """Entry point for the ``fp`` command."""
+    _check_python_version()
+
+    import sys as _sys
+
+    if argv is None:
+        argv = _sys.argv[1:]
+    from footprinter.paths import is_test_mode
+    from footprinter.source_registry import ConfigError as _ConfigError
+
+    if is_test_mode():
+        print(
+            "\033[33m⚠ Test mode active — data at ~/.footprinter/.test-env/\033[0m",
+            file=_sys.stderr,
+        )
+
+    from footprinter import __version__
+    from footprinter.cli._common import FORMATTER
+
+    parser = argparse.ArgumentParser(
+        prog="fp",
+        description=f"Footprinter v{__version__} — file archival and AI context CLI",
+        epilog=(
+            "getting started:\n"
+            "  fp setup                   Run the configuration wizard\n"
+            "  fp setup --check           Validate existing configuration\n"
+            "  fp connect list            Show available data source connectors\n"
+            "\n"
+            "data commands:\n"
+            "  fp ingest                  Run the data ingest pipeline (incremental)\n"
+            "  fp ingest --full           Re-process all data sources\n"
+            "  fp status                  Show data counts and system health\n"
+            "  fp search 'my query'       Semantic search across indexed content\n"
+            "\n"
+            "browse indexed data:\n"
+            "  fp view files               List indexed files\n"
+            "  fp view folders             List indexed folders\n"
+            "  fp view projects            List projects\n"
+            "  fp view clients             List clients\n"
+            "  fp view chats               List chats\n"
+            "  fp view emails              List indexed emails\n"
+            "  fp view visits              List browser history\n"
+            "\n"
+            "servers:\n"
+            "  fp mcp                     Start the MCP server\n"
+            "  fp api                     Start the HTTP API server\n"
+            "  fp mcp check ~/Work/file   Check access resolution for a path\n"
+            "\n"
+            "tip: use 'fp <command> --help' for details on any command."
+        ),
+        formatter_class=FORMATTER,
+    )
+    parser.add_argument(
+        "--version",
+        action="version",
+        version=f"%(prog)s {__version__}",
+    )
+    subparsers = parser.add_subparsers(dest="subcommand", metavar="COMMAND")
+    subparsers.required = False
+
+    from footprinter.cli import (
+        api_cmd,
+        connect,
+        data,
+        delete,
+        ingest,
+        mcp_cmd,
+        search_cmd,
+        setup,
+        status_cmd,
+        upsert,
+        vectorize_cmd,
+        view,
+    )
+
+    for mod in [
+        ingest,
+        mcp_cmd,
+        api_cmd,
+        status_cmd,
+        search_cmd,
+        setup,
+        connect,
+        view,
+        upsert,
+        data,
+        delete,
+        vectorize_cmd,
+    ]:
+        mod.register(subparsers)
+
+    args = parser.parse_args(argv)
+    if args.subcommand is None:
+        if _is_first_run():
+            print(
+                "\033[33;1m\U0001f4a1 Looks like this is your first time running "
+                "Footprinter.\n   Run 'fp setup' to get started.\033[0m\n",
+                file=_sys.stderr,
+            )
+        parser.print_help()
+        return
+    from footprinter.cli._prompt import PromptCancelled
+
+    try:
+        args.func(args)
+    except _ConfigError as e:
+        print(str(e), file=_sys.stderr)
+        _sys.exit(1)
+    except (PromptCancelled, KeyboardInterrupt):
+        print("\nCancelled.", file=_sys.stderr)
+        _sys.exit(130)

footprinter/cli/__main__.py

@@ -0,0 +1,6 @@
+"""Allow ``python -m footprinter.cli`` invocation."""
+
+from footprinter.cli import main
+
+if __name__ == "__main__":
+    main()