Skip to content

Supported Formats

Jean-Baptiste THERY edited this page Nov 15, 2026 · 5 revisions

Supported formats

Ragmir indexes every format below with no extra install, unless the table says otherwise. Each result carries a citation an agent can quote and open again: lines for text, pages for PDF, slides, sheets and cells, EPUB spine items, a passage number for reformatted text, or a named section for one object of a mailbox, design file, data model, diagram, API description, or collection. rgr audit --unsupported lists the files of your sources that Ragmir skipped.

Every format

Family Extensions and file names Citation
Documents .pdf, .docx, .docm, .dotx, .dotm, .pptx, .pptm, .potx, .potm, .ppsx, .ppsm, .xlsx, .xlsm, .xltx, .xltm, .odt, .ods, .odp, .odg, .ott, .ots, .otp, .otg, .fodt, .fods, .fodp, .fodg, .epub, .fb2, .rtf, .html, .htm, .xhtml, .xht, .shtml, .mhtml, .mht Page, slide, sheet and cells, EPUB item, chapter, or passage
Writing and markup .md, .markdown, .mdown, .mkd, .mkdn, .mdwn, .mdtext, .mdtxt, .mdx, .mdoc, .mdc, .livemd, .qmd, .rmd, .rnw, .txt, .text, .adoc, .asciidoc, .rst, .org, .typ, .tex, .ltx, .sty, .cls, .bib, .ris, .textile, .wiki, .mediawiki, .creole, .rdoc, .pod, .texi, .texinfo, .txi, .opml, .wikitext, .pod6, .scrbl, .rd, .bibtex, .bst, .dtx, .ins, .mkiv, .dita, .ditamap, .ditaval, .man, .roff, .ronn Lines
Mail, contacts, calendar, captions, and feeds .eml, .emlx, .mbox, .msg, .vcf, .ics, .ical, .vcs, .srt, .vtt, .ass, .ssa, .sbv, .lrc, .ttml, .dfxp, .rss, .atom Lines; .eml, .emlx, .msg: passage; .mbox: one section per message
Notes, whiteboards, mind maps, and design files .enex, .canvas, .excalidraw, .tldr, .fig, .sketch, .xmind Section per note, frame, page, group, artboard, or mind map sheet
Data and configuration .csv, .tsv, .json, .jsonl, .ndjson, .json5, .jsonc, .hjson, .cson, .jsonld, .geojson, .topojson, .webmanifest, .ipynb, .yaml, .yml, .toml, .ini, .cfg, .conf, .cnf, .properties, .xml, .log, .example, .exemple, .sample, .template, .dist, .desktop, .service, .timer, .socket, .mount, .automount, .target, .slice, .container, .volume, .network, .kube, .caddyfile, .apacheconf, .vhost, .vcl, .xcconfig, .entitlements, .plist, .code-workspace, .code-snippets, .hujson, .ron, .textproto, .pbtxt, .zon, .urdf, .xacro, .launch, .kdl, .hocon Lines (JSON and YAML: passage; GeoJSON: section per feature)
Translations .po, .pot, .xliff, .xlf, .strings, .stringsdict, .resx, .resw, .arb, .ftl, .xcstrings Lines
Data models, SQL, and queries .cdm, .ldm, .pdm, .oom, .bpm, .mcd, .mpd, .mwb, .dmd, .sql, .psql, .pgsql, .mysql, .ddl, .dml, .pkb, .pks, .pls, .plb, .pck, .bdy, .trg, .fnc, .prc, .dbml, .prisma, .hql, .cql, .cypher, .sparql, .rq, .ttl, .rdf, .owl, .edgeql, .surql, .kql, .prql, .plsql, .cyp, .esdl, .cds, .asddls, .xquery, .xq, .xqm Section per object, or lines
Diagrams and processes .drawio, .dio, .vsdx, .vsdm, .vstx, .vstm, .bpmn, .dmn, .archimate, .xmi, .mmd, .mermaid, .puml, .plantuml, .pu, .iuml, .wsd, .dsl, .d2, .dot, .gv, .graphml, .likec4, .c4, .svg, .scxml Section per page or object, or lines
APIs, schemas, and requests .graphql, .gql, .proto, .avsc, .thrift, .xsd, .wsdl, .http, .rest, .hurl, .bru, .har, .raml, .smithy, .tsp, .capnp, .fbs, .idl, .wit, .avdl, .graphqls, .apib, .webidl, .aidl Section per operation or request, or lines
BI, reporting, analytics, and ETL .twb, .twbx, .tmdl, .pbip, .pbir, .pbism, .bim, .jrxml, .lkml, .malloy, .dax, .pq, .qvs, .rdl, .rdlc, .dtsx, .ktr, .kjb, .yxmd, .yxmc, .yxwz, .sas, .sps, .do, .ado Section per object, or lines
Code .ts, .tsx, .mts, .cts, .js, .jsx, .mjs, .cjs, .vue, .svelte, .astro, .marko, .css, .scss, .sass, .less, .styl, .pcss, .postcss, .py, .pyi, .pyx, .pxd, .java, .kt, .kts, .scala, .sbt, .groovy, .gradle, .go, .rs, .zig, .nim, .c, .h, .cc, .cpp, .cxx, .hh, .hpp, .hxx, .ino, .m, .mm, .cu, .cuh, .cs, .vb, .fs, .fsx, .fsi, .php, .rb, .rake, .gemspec, .podspec, .swift, .dart, .lua, .r, .jl, .pl, .ex, .exs, .erl, .gleam, .hs, .cabal, .ml, .mli, .re, .sml, .elm, .purs, .clj, .cljs, .cljc, .edn, .lisp, .lsp, .el, .scm, .ss, .rkt, .cr, .hx, .mojo, .v, .sv, .svh, .vhd, .vhdl, .f, .for, .f77, .f90, .f95, .f03, .f08, .pas, .pp, .ada, .adb, .ads, .asm, .s, .wat, .glsl, .vert, .frag, .hlsl, .wgsl, .metal, .sol, .move, .cairo, .vy, .cbl, .cob, .cpy, .jcl, .pli, .rpgle, .abap, .trigger, .cfm, .cfc, .bas, .vbs, .tcl, .coffee, .wl, .wls, .qml, .xaml, .axaml, .xib, .storyboard, .ui, .vim, .ets, .bal, .cdc, .carbon, .odin, .k, .slint, .hrl, .lhs, .mlir, .dm, .as, .al, .ceylon, .chpl, .roc, .lean, .agda, .idr, .tla, .smt2, .dfy, .fnl, .janet, .hy, .pony, .vala, .ls, .moon, .wren, .gd, .gdshader, .shader, .tres, .tscn, .uc, .nut, .luau, .au3, .nu, .dpr, .lpr, .apex, .stan, .fxml, .mxml, .glade, .xul, .c++, .h++, .cppm, .ixx, .inl, .ipp, .tpp, .tcc, .txx, .csx, .cake, .linq, .jsh, .gvy, .gtpl, .sc, .es6, .jsm, .gs, .gjs, .gts, .pm, .t, .psgi, .raku, .rakumod, .pm6, .phpt, .pyw, .gyp, .gypi, .wsgi, .rpy, .spec, .rbi, .ru, .jbuilder, .builder, .rabl, .thor, .bats, .command, .xsh, .elv, .rockspec, .nse, .escript, .xrl, .yrl, .hsc, .mll, .mly, .bb, .cl, .sld, .nimble, .nims, .vapi, .pxi, .hack, .rei, .ah2, .tm, .cobol, .rexx, .4gl, .pc, .sqlrpgle, .bsl, .prolog, .qbs, .ncl, .geom, .tesc, .tese, .comp, .rgen, .rchit, .rmiss, .vsh, .fsh, .hlsli, .fx, .fxh, .cginc, .gdshaderinc, .wesl, .slang, .hip, .ispc, .vh, .xdc, .fpp, .mata, .sthlp, .nf, .smk, .cwl, .wdl, .sage, .gnuplot, .thy, .pde, .scad, .qs, .qasm, .circom, .nr, .clar, .cj, .c3, .ha, .mbt, .jai, .imba, .yul, .tact, .ink, .feature, .robot, .g4, .y, .l, .bison, .pegjs, .peggy, .jq, .ql, .qll, .xsl, .xslt, .xspec, .yar, .yara, .zeek, .sieve Lines
Web templates .hbs, .handlebars, .mustache, .ejs, .erb, .haml, .slim, .pug, .jade, .njk, .jinja, .jinja2, .j2, .twig, .liquid, .cshtml, .vbhtml, .razor, .jsp, .jspx, .asp, .aspx, .ascx, .gohtml, .gotmpl, .tmpl, .mjml, .templ, .phtml, .ctp, .eex, .heex, .leex, .gsp, .edge, .latte, .vtl, .jte, .mako, .soy, .riot, .svx, .vto Lines
Infrastructure, build, and scripts .tf, .hcl, .tpl, .bicep, .bicepparam, .nomad, .rego, .sentinel, .cue, .jsonnet, .libsonnet, .nix, .dhall, .pkl, .kcl, .sls, .dockerfile, .containerfile, .cmake, .bzl, .bazel, .star, .mk, .mak, .ninja, .gn, .gni, .sh, .bash, .zsh, .fish, .ksh, .csh, .tcsh, .ps1, .psm1, .psd1, .bat, .cmd, .awk, .sed, .ahk, .applescript, .diff, .patch, .csproj, .fsproj, .vbproj, .vcxproj, .sqlproj, .shproj, .wixproj, .proj, .props, .targets, .nuspec, .sln, .slnx, .natvis, .vsixmanifest, .vstemplate, .ps1xml, .wxs, .wxi, .wxl, .iss, .nsi, .nsh, .tofu, .tftpl, .make, .just, .ld, .td, .ebuild, .eclass, .bbappend, .bbclass Lines
Security and compliance reports .sarif, .spdx Section per finding (SARIF), or lines
Project files without an extension AUTHORS, CHANGELOG, CHANGES, CITATION, CITATION.cff, CODEOWNERS, CONTRIBUTORS, COPYING, HISTORY, LICENCE, LICENSE, NEWS, NOTICE, OWNERS, README, TODO Lines
Build and automation files Appfile, Berksfile, Brewfile, BUILD, Caddyfile, Capfile, Cartfile, Containerfile, Dangerfile, Dockerfile, Earthfile, Fastfile, Gemfile, GNUmakefile, Gradlew, Guardfile, Jenkinsfile, Justfile, Makefile, Matchfile, Mvnw, Pipfile, Podfile, Procfile, Rakefile, Snakefile, Tiltfile, Vagrantfile, WORKSPACE, and variants such as Dockerfile.*, Containerfile.*, Jenkinsfile.*, Makefile.*, Android.bp, APKBUILD, Appraisals, BSDmakefile, BUCK, Buildfile, cabal.project, Cakefile, Cask, configure.ac, cpanfile, Deliverfile, DEPS, dune-project, Eask, Emakefile, firestore.rules, go.mod, go.work, Jakefile, Jarfile, Kbuild, kcl.mod, MANIFEST.MF, Mavenfile, meson.build, mkfile, Modulefile, nextflow.config, Phakefile, PKGBUILD, project.godot, Puppetfile, rebar.config, Rexfile, SConscript, SConstruct, Singularity, Snapfile, Steepfile, Thorfile, wscript Lines
Tool configuration files .babelrc, .browserslistrc, .clang-format, .clang-tidy, .clangd, .containerignore, .dockerignore, .editorconfig, .eslintignore, .eslintrc, .flake8, .gcloudignore, .gitattributes, .gitignore, .gitmodules, .helmignore, .htaccess, .mailmap, .node-version, .npmignore, .nvmrc, .prettierignore, .prettierrc, .pylintrc, .python-version, .ruby-version, .stylelintignore, .stylelintrc, .swcrc, .terraform-version, .tool-versions, .vscodeignore, .all-contributorsrc, .arcconfig, .autorc, .buckconfig, .c8rc, .codespellrc, .coveragerc, .gitreview, .htmlhintrc, .irbrc, .jscsrc, .jshintrc, .jslintrc, .justfile, .latexmkrc, .luacheckrc, .nycrc, .pryrc, .releaserc, .Rprofile, .rspec, .shellcheckrc, .simplecov, .tern-config, .tern-project, .watchmanconfig, .yardopts, _redirects, browserslist, latexmkrc, mocha.opts, pylintrc Lines
With a local extractor (optional) Scanned PDF pages; images .png, .jpg, .jpeg, .tif, .tiff, .webp, .gif, .bmp, .heic, .heif, .avif; legacy Word .doc Page or passage

Dotenv templates such as .env.example, .env.sample, or env.template keep their variable names and comments, each on its own line, with every value masked; .env and every other .env.* file stay skipped as secrets, like Terraform state and variable files (.tfstate, .tfvars) and the Firebase GoogleService-Info.plist.

Office charts are indexed with their titles, axes, series, and values. Text files keep their own encoding: a UTF-16 byte order mark selects UTF-16, and bytes that are not valid UTF-8 are read as Windows-1252, so legacy CSV files, subtitles, and Apple string tables index correctly. Other text extensions can be added with includeExtensions in the configuration. Binary formats outside this list are reported, never guessed.

Structured exports in detail

Tool or standard Files What each section holds
PowerDesigner .cdm, .ldm, .pdm, .oom, .bpm, .mcd, .mpd, or .xml saved as XML Entities, tables, views, associations, references, inheritances, domains, business rules, classes, and packages, with columns or attributes, types, comments, keys, indexes, joins, and the objects that reference them
ERwin Data Modeler XML export (.xml) Entities, attributes with physical names and types, relationships
MySQL Workbench .mwb Tables, columns, foreign keys, indexes, views, routines
Oracle SQL Developer Data Modeler object .xml files in the design folder Entities, tables, and their attributes or columns with comments
draw.io / diagrams.net .drawio, .dio, .drawio.svg, or .xml One section per page: shape labels, custom properties, containers, and connections
Visio .vsdx, .vsdm, .vstx, .vstm One section per page: shape text and connectors
BPMN 2.0 .bpmn or .xml Processes with lanes, steps, documentation, flows, and conditions; collaborations
DMN .dmn or .xml Decisions with hit policy, inputs, outputs, every rule, and requirements
ArchiMate .archimate (Archi) or Open Group exchange .xml Elements with documentation, properties, and relations
UML .xmi or .xml Classes, interfaces, enumerations, components, actors, use cases, attributes, operations, generalizations, notes
Figma, FigJam, Figma Slides .fig saved with "Save local copy", or the file JSON of the REST API One section per top-level frame, section, component, or slide of each page: text layers, instance text overrides, component descriptions; loose stickies and text per page
Sketch .sketch One section per artboard or symbol: text layers and symbol text overrides
Excalidraw .excalidraw, or a .json scene One section per frame and one for the rest of the canvas: text, labelled shapes, labelled arrows, links
tldraw .tldr One section per page: frame names, text and notes in reading order, bound arrows
JSON Canvas (Obsidian) .canvas One section per group and one for ungrouped cards: text, file and link cards, labelled edges
XMind .xmind from XMind 8, or XMind 2020 and later One section per sheet: the topic outline with notes, labels, task and priority markers, web links, floating, callout, and summary topics, and labelled relationships
Tableau .twb, .twbx Data sources with fields and calculated formulas, worksheets, dashboards
OpenAPI, Swagger, AsyncAPI .yaml, .yml, .json API overview, each operation (GET /invoices/{id}), channel, message, and schema
Postman collection v2.0 or v2.1 .json Collection overview, folders, and one section per request: method, URL, description, parameters, headers, body, example responses; secret header values left out
Insomnia v4 .json export or v5 .yaml collection Workspace, then one section per request with its folder path; environments are not indexed
HTTP Archive (HAR) .har saved from browser DevTools, Charles, Proxyman, or Fiddler An overview, then one section per API or page request: method, URL, status, query parameters, headers, and text bodies, with identical repeated requests counted once; cookies, credentials, tokens, and secret-like parameters left out, and static assets and CORS preflights skipped
Power BI project .tmdl, .pbip, .pbir, .pbism, .bim TMDL as text with line citations, PBIR and TMSL as JSON
Email .eml, .emlx, .msg Subject, sender, recipients, date, attachment names, and the decoded body: quoted-printable and base64 parts in their charset, HTML rendered as text, forwarded messages included
Mailbox .mbox (Gmail Takeout, Thunderbird, Apple Mail export) One section per message, labelled with its number and subject
Saved web page .mhtml, .mht Page title, address, and the rendered page; images are skipped
Evernote .enex One section per note: title, tags, dates, author, source address, attachment names, and the rendered body
FictionBook .fb2 Book description, then one section per chapter
GeoJSON .geojson, or a .json feature collection One section per feature: its name, geometry type, point coordinates, and properties
SARIF .sarif from CodeQL, Semgrep, ESLint, and other analyzers One section per finding: rule, level, message, locations, and rule description
Flat OpenDocument .fodt, .fods, .fodp, .fodg Metadata and body, without styles and settings

How detection works

A file with a dedicated extension uses its extractor. A generic .xml file is read as structured only when its first bytes identify the tool: the PowerDesigner processing instruction, or the root element and namespace of draw.io, BPMN, DMN, ArchiMate, XMI, MySQL Workbench, Oracle Data Modeler, ERwin, or Tableau. Any other XML file, such as a Maven pom.xml, keeps line citations. A .json or .yaml file is read as structured only when its shape identifies an OpenAPI or AsyncAPI description, a Postman or Insomnia collection, a Figma file, an Excalidraw or tldraw document, a GeoJSON feature collection, a SARIF log, or a JSON Canvas; any other JSON keeps its plain text.

Exports that need one step first

  • PowerDesigner binary models. Save the model in XML format. Ragmir reports a binary model as a file error and keeps indexing the rest of the corpus. PowerDesigner backups (.bcd, .bpd) are not indexed because they duplicate the model.
  • Oracle SQL Developer Data Modeler. The .dmd file only points to the design folder. Add the folder next to it to your sources so the object files are indexed.
  • Power BI .pbix. The model is stored in a compressed binary part. Save the report as a Power BI project (PBIP) with the TMDL format and index the project folder.
  • Enterprise Architect .qea and .eap. These are databases. Export the model to XMI.
  • Figma in the browser. Use "Save local copy" to get a .fig file, or save the file JSON of the REST API (GET /v1/files/:key). Zstandard-compressed .fig files need Node.js 22.15 or later.
  • Apple Pages, Numbers, and Keynote. Export to Word, Excel, PowerPoint, or PDF.
  • OneNote, Google Docs, Notion, and Confluence. Export to Word, PDF, HTML, or Markdown; Google Takeout and Notion exports already use these formats.
  • Legacy .xls and .ppt, and binary .xlsb workbooks. Convert to .xlsx, .pptx, CSV, or PDF.
  • Binary property lists. Xcode writes .plist files as XML, which index with line citations. A binary one is reported as a file error: save an XML copy with plutil -convert xml1 -o copy.plist <file> and index the copy.
  • Password-protected XMind maps. Save a copy without a password; Ragmir reports the protected map as a file error.

Limits

One file produces at most 20,000 sections; a last section records how many objects were left out. A compressed draw.io page expands to at most 25 MB, and a Figma document to at most 256 MB. Archives (.mwb, .vsdx, .twbx, .sketch, .xmind) use the same entry-count and size limits as Office files, and a mind map is read 64 topic levels deep, with deeper branches counted. A mail part is indexed up to 2 million characters; binary attachments are listed by name. Files larger than maxFileBytes (50 MB at most) are skipped, so split large mailboxes by year or label before indexing them. A damaged file is reported in the ingestion errors and does not stop ingestion.

After an upgrade, the next rgr ingest stages a rebuild with the new extractors and activates it only once it passes validation. The previous index stays searchable meanwhile, and .ragmir/storage/ never needs to be deleted.

Clone this wiki locally