text-mirror 0.7.0

# Format table for text-mirror.
#
# Detection maps magic bytes and file extensions to the format ids
# below. The version is recorded in every manifest record as
# rules_version, so a change here is a visible versioned event. It must
# match the version in converters.toml.

version = "10"

[[formats]]
id = "text"
name = "Plain text"
extensions = ["txt", "text", "log"]

[[formats]]
id = "markdown"
name = "Markdown"
extensions = ["md", "markdown"]

[[formats]]
id = "html"
name = "HTML"
extensions = ["html", "htm", "xhtml"]

[[formats]]
id = "csv"
name = "Comma-separated values"
extensions = ["csv"]

[[formats]]
id = "rtf"
name = "Rich Text Format"
extensions = ["rtf"]

[[formats]]
id = "odt"
name = "OpenDocument text"
extensions = ["odt"]

[[formats]]
id = "ods"
name = "OpenDocument spreadsheet"
extensions = ["ods"]

[[formats]]
id = "odp"
name = "OpenDocument presentation"
extensions = ["odp"]

[[formats]]
id = "epub"
name = "EPUB ebook"
extensions = ["epub"]

[[formats]]
id = "docx"
name = "Office Open XML document"
extensions = ["docx"]

[[formats]]
id = "xlsx"
name = "Office Open XML spreadsheet"
extensions = ["xlsx"]

[[formats]]
id = "xlsm"
name = "Office Open XML macro-enabled spreadsheet"
extensions = ["xlsm"]

[[formats]]
id = "xlsb"
name = "Excel binary spreadsheet"
extensions = ["xlsb"]

[[formats]]
id = "xla"
name = "Excel legacy add-in"
extensions = ["xla"]

[[formats]]
id = "xlam"
name = "Excel macro-enabled add-in"
extensions = ["xlam"]

[[formats]]
id = "pptx"
name = "Office Open XML presentation"
extensions = ["pptx"]

[[formats]]
id = "doc"
name = "Legacy Word document"
extensions = ["doc"]

[[formats]]
id = "xls"
name = "Legacy Excel spreadsheet"
extensions = ["xls"]

[[formats]]
id = "ppt"
name = "Legacy PowerPoint presentation"
extensions = ["ppt"]

[[formats]]
id = "pdf"
name = "Portable Document Format"
extensions = ["pdf"]

# Modern Adobe Illustrator artwork is a PDF-compatible container, so
# magic resolves the bytes to the generic pdf parent and the declared
# .ai extension refines it to this more specific id with no mismatch,
# exactly as svg refines xml. A PDF-magic file with no .ai extension
# stays pdf. A legacy PostScript-backed .ai, whose magic the table
# does not name, resolves by extension to ai.
[[formats]]
id = "ai"
name = "Adobe Illustrator artwork"
extensions = ["ai"]
refines = "pdf"

[[formats]]
id = "png"
name = "Portable Network Graphics image"
extensions = ["png"]

[[formats]]
id = "jpeg"
name = "JPEG image"
extensions = ["jpg", "jpeg"]

[[formats]]
id = "tiff"
name = "TIFF image"
extensions = ["tif", "tiff"]

[[formats]]
id = "mp3"
name = "MP3 audio"
extensions = ["mp3"]

[[formats]]
id = "wav"
name = "WAV audio"
extensions = ["wav"]

[[formats]]
id = "m4a"
name = "MPEG-4 audio"
extensions = ["m4a"]

[[formats]]
id = "flac"
name = "FLAC audio"
extensions = ["flac"]

[[formats]]
id = "mp4"
name = "MPEG-4 video"
extensions = ["mp4"]

[[formats]]
id = "mov"
name = "QuickTime video"
extensions = ["mov"]

[[formats]]
id = "zip"
name = "ZIP archive"
extensions = ["zip"]

[[formats]]
id = "eml"
name = "Email message"
extensions = ["eml"]

[[formats]]
id = "msg"
name = "Outlook message"
extensions = ["msg"]

# Text-native formats. Each routes to the plain text passthrough in
# converters.toml, which validates UTF-8 and fails closed on anything
# else, so a binary wearing one of these extensions records failed
# with invalid_utf8. One id per format so consumers can slice by
# detected_format.

[[formats]]
id = "json"
name = "JSON"
extensions = ["json"]

[[formats]]
id = "jsonl"
name = "JSON Lines"
extensions = ["jsonl", "ndjson"]

[[formats]]
id = "yaml"
name = "YAML"
extensions = ["yaml", "yml"]

[[formats]]
id = "xml"
name = "XML"
extensions = ["xml"]

[[formats]]
id = "tsv"
name = "Tab-separated values"
extensions = ["tsv"]

[[formats]]
id = "toml"
name = "TOML"
extensions = ["toml"]

[[formats]]
id = "ini"
name = "INI configuration"
extensions = ["ini", "cfg", "conf", "properties"]

# basenames matches exact file names for paths with no extension, so
# a file named exactly .env still declares its format.
[[formats]]
id = "env"
name = "Environment file"
extensions = ["env"]
basenames = [".env"]

# refines names the generic parent format whose magic these bytes
# also satisfy. When magic resolves to the parent and the extension
# declares the refinement, the declared id wins with no mismatch,
# because an svg is xml and the observations agree.
[[formats]]
id = "svg"
name = "Scalable Vector Graphics"
extensions = ["svg"]
refines = "xml"

[[formats]]
id = "ipynb"
name = "Jupyter notebook"
extensions = ["ipynb"]

[[formats]]
id = "python"
name = "Python source"
extensions = ["py"]

[[formats]]
id = "shell"
name = "Shell script"
extensions = ["sh", "bash", "zsh"]

[[formats]]
id = "rust"
name = "Rust source"
extensions = ["rs"]

[[formats]]
id = "javascript"
name = "JavaScript source"
extensions = ["js", "mjs", "jsx"]

[[formats]]
id = "typescript"
name = "TypeScript source"
extensions = ["ts", "tsx"]

[[formats]]
id = "sql"
name = "SQL script"
extensions = ["sql"]

[[formats]]
id = "ruby"
name = "Ruby source"
extensions = ["rb"]

[[formats]]
id = "go"
name = "Go source"
extensions = ["go"]

[[formats]]
id = "java"
name = "Java source"
extensions = ["java"]

[[formats]]
id = "c"
name = "C source"
extensions = ["c", "h"]

[[formats]]
id = "cpp"
name = "C++ source"
extensions = ["cc", "cpp", "hpp"]

[[formats]]
id = "csharp"
name = "C# source"
extensions = ["cs"]

[[formats]]
id = "php"
name = "PHP source"
extensions = ["php"]

[[formats]]
id = "perl"
name = "Perl source"
extensions = ["pl"]

[[formats]]
id = "powershell"
name = "PowerShell script"
extensions = ["ps1", "psm1"]

[[formats]]
id = "batch"
name = "Windows batch script"
extensions = ["bat", "cmd"]

[[formats]]
id = "terraform"
name = "Terraform configuration"
extensions = ["tf", "tfvars"]

[[formats]]
id = "hcl"
name = "HashiCorp configuration language"
extensions = ["hcl"]

[[formats]]
id = "lookml"
name = "LookML model"
extensions = ["lkml"]

[[formats]]
id = "tmdl"
name = "Tabular Model Definition Language"
extensions = ["tmdl"]

[[formats]]
id = "pbip"
name = "Power BI project"
extensions = ["pbip"]

[[formats]]
id = "tableau-workbook"
name = "Tableau workbook"
extensions = ["twb"]
refines = "xml"

[[formats]]
id = "tableau-datasource"
name = "Tableau data source"
extensions = ["tds"]
refines = "xml"

# Recognized formats no converter claims. Each carries a declared
# unsupported reason in converters.toml, so detection names the
# format and the record names the reason.

[[formats]]
id = "pickle"
name = "Python pickle"
extensions = ["pkl", "pickle"]

[[formats]]
id = "webp"
name = "WebP image"
extensions = ["webp"]

[[formats]]
id = "heic"
name = "HEIC image"
extensions = ["heic"]

[[formats]]
id = "parquet"
name = "Apache Parquet"
extensions = ["parquet"]

[[formats]]
id = "avro"
name = "Apache Avro"
extensions = ["avro"]

[[formats]]
id = "sqlite"
name = "SQLite database"
extensions = ["sqlite", "sqlite3"]

[[formats]]
id = "arrow"
name = "Apache Arrow"
extensions = ["arrow"]

[[formats]]
id = "orc"
name = "Apache ORC"
extensions = ["orc"]

[[formats]]
id = "feather"
name = "Feather columnar"
extensions = ["feather"]

[[formats]]
id = "duckdb"
name = "DuckDB database"
extensions = ["duckdb"]

[[formats]]
id = "faiss"
name = "FAISS index"
extensions = ["faiss"]

[[formats]]
id = "msgpack"
name = "MessagePack data"
extensions = ["msgpack"]

[[formats]]
id = "tar"
name = "Tar archive"
extensions = ["tar"]

[[formats]]
id = "gzip"
name = "Gzip compressed file"
extensions = ["gz"]

[[formats]]
id = "bzip2"
name = "Bzip2 compressed file"
extensions = ["bz2"]

[[formats]]
id = "7z"
name = "7-Zip archive"
extensions = ["7z"]

[[formats]]
id = "rar"
name = "RAR archive"
extensions = ["rar"]

[[formats]]
id = "sketch"
name = "Sketch design"
extensions = ["sketch"]

[[formats]]
id = "psd"
name = "Adobe Photoshop document"
extensions = ["psd", "psb"]

[[formats]]
id = "fig"
name = "Figma design"
extensions = ["fig"]

[[formats]]
id = "indd"
name = "Adobe InDesign document"
extensions = ["indd"]

[[formats]]
id = "idml"
name = "Adobe InDesign markup"
extensions = ["idml"]