1 ! Copyright (C) 2022 Doug Coleman.
2 ! See https://factorcode.org/license.txt for BSD license.
3 USING: accessors assocs assocs.extras combinators
4 combinators.short-circuit formatting hashtables io io.backend
5 io.directories io.encodings.binary io.files io.files.info
6 io.files.types io.pathnames kernel math math.statistics
7 prettyprint sequences sets sorting splitting toml
8 tools.memory.private tools.wc unicode ;
11 : file-sizes ( paths -- assoc )
12 [ dup file-info size>> ] { } map>assoc ;
14 : binary-file? ( path -- ? )
15 binary [ 1024 read ] with-file-reader [ 0 = ] any? ;
17 : binary-files ( paths -- ? ) [ binary-file? ] filter ;
19 : partition-binary ( paths -- binary text )
20 [ binary-file? ] partition ;
22 : with-file-extensions ( paths -- paths' )
23 [ has-file-extension? ] filter ;
25 : without-git-paths ( paths -- paths' )
26 [ "/.git/" subseq-of? ] reject ;
28 : without-node-modules-paths ( paths -- paths' )
29 [ "/node_modules/" subseq-of? ] reject ;
31 : regular-directory-files ( path -- seq )
32 recursive-directory-files
33 [ link-info type>> +regular-file+ = ] filter ;
35 : codebase-paths ( path -- seq )
36 regular-directory-files
38 without-node-modules-paths ;
40 : count-by-file-extension ( paths -- assoc )
42 [ file-extension ] histogram-by
45 : collect-extensions-by-line-count ( paths -- assoc )
50 : collect-by-file-extension ( paths -- assoc )
52 [ file-extension ] collect-by ;
54 : sum-line-counts-by-extension ( paths -- assoc )
55 [ binary-file? ] reject
56 collect-by-file-extension
57 [ [ wc ] map-sum ] assoc-map
60 : sum-sizes-by-extension ( paths -- assoc )
61 collect-by-file-extension
62 [ [ file-info size>> ] map-sum ] assoc-map
65 : upper-file? ( path -- ? )
67 [ { [ file-stem length 4 > ] [ file-extension length 3 <= ] } 1&& ]
69 [ file-stem [ { [ digit? ] [ "-." member? ] } 1|| ] all? not ]
71 : upper-files ( paths -- seq ) [ upper-file? ] filter ;
73 : configure-file? ( path -- ? ) file-name >lower { [ "configure" = ] [ "configure." head? ] } 1|| ;
74 : configure-files ( paths -- paths' ) [ configure-file? ] filter ;
76 : cmake-file? ( path -- ? ) { [ "CMakeLists.txt" tail? ] [ ".cmake" tail? ] } 1|| ;
77 : cmake-files ( paths -- paths' ) [ cmake-file? ] filter ;
78 : uses-cmake? ( paths -- ? ) [ cmake-file? ] any? ;
80 : in-file? ( paths -- ? ) ".in" tail? ;
81 : in-files ( paths -- seq ) [ in-file? ] filter ;
82 : uses-in-files? ( paths -- ? ) [ in-file? ] any? ;
84 : shell-file? ( path -- ? ) >lower file-extension { "sh" "zsh" } member? ;
85 : shell-files ( paths -- paths' ) [ shell-file? ] filter ;
86 : uses-shell? ( paths -- ? ) [ shell-file? ] any? ;
88 : swift-files ( paths -- paths' ) [ ".swift" tail? ] filter ;
90 : c-file? ( path -- ? )
91 >lower file-extension { "c" } member? ;
92 : c-files ( paths -- paths' ) [ c-file? ] filter ;
94 : c-header-file? ( path -- ? )
95 >lower file-extension { "h" } member? ;
96 : c-header-files ( paths -- paths' ) [ c-header-file? ] filter ;
98 : cpp-file? ( path -- ? )
99 >lower file-extension { "cc" "cpp" } member? ;
100 : cpp-files ( paths -- paths' ) [ cpp-file? ] filter ;
102 : cpp-header-file? ( path -- ? )
103 >lower file-extension { "h" "hh" "hpp" } member? ;
104 : cpp-header-files ( paths -- paths' ) [ cpp-header-file? ] filter ;
106 : python-file? ( path -- ? )
107 >lower file-extension {
108 "py" "py3" "pyc" "pyo" "pyw" "pyx" "pyd"
109 "pxd" "pxi" "pyd" "pxi" "pyi" "pyz" "pwxz" "pth"
111 : python-files ( paths -- paths' ) [ python-file? ] filter ;
113 : markdown-file? ( path -- ? ) { [ ".md" tail? ] [ ".markdown" tail? ] } 1|| ;
114 : markdown-files ( paths -- paths' ) [ markdown-file? ] filter ;
116 : dot-file? ( path -- ? ) file-name "." head? ;
117 : dot-files ( paths -- paths' ) [ dot-file? ] filter ;
119 : txt-file? ( path -- ? )
121 [ { [ ".txt" tail? ] [ ".TXT" tail? ] } 1|| ]
122 [ "CMakeLists.txt" tail? not ]
124 : txt-files ( paths -- paths' ) [ txt-file? ] filter ;
126 : license-file? ( path -- ? )
127 >lower { [ file-stem "license" = ] [ "license-mit" tail? ] } 1|| ;
128 : license-files ( paths -- paths' ) [ license-file? ] filter ;
130 : readme-file? ( path -- ? )
131 >lower file-stem "readme" = ;
132 : readme-files ( paths -- paths' ) [ readme-file? ] filter ;
134 : owners-file? ( path -- ? )
135 >lower file-stem "owners" = ;
136 : owners-files ( paths -- paths' ) [ owners-file? ] filter ;
138 : codenotify-file? ( path -- ? )
139 >lower file-stem "codenotify" = ;
140 : codenotify-files ( paths -- paths' ) [ codenotify-file? ] filter ;
142 : contributing-file? ( path -- ? )
143 >lower file-stem "contributing" = ;
144 : contributing-files ( paths -- paths' ) [ contributing-file? ] filter ;
146 : changelog-file? ( path -- ? )
147 >lower file-stem "changelog" = ;
148 : changelog-files ( paths -- paths' ) [ changelog-file? ] filter ;
150 : security-file? ( path -- ? )
151 >lower file-stem "security" = ;
152 : security-files ( paths -- paths' ) [ security-file? ] filter ;
154 : notice-file? ( path -- ? )
155 >lower file-stem "notice" = ;
156 : notice-files ( paths -- paths' ) [ notice-file? ] filter ;
158 : version-file? ( path -- ? )
159 >lower file-stem "version" = ;
160 : version-files ( paths -- paths' ) [ version-file? ] filter ;
162 : json-file? ( path -- ? )
163 >lower file-extension
164 { "json" "jsonc" } member? ;
166 : json-files ( paths -- paths' ) [ json-file? ] filter ;
168 : yaml-file? ( path -- ? ) { [ ".yaml" tail? ] [ ".yml" tail? ] } 1|| ;
169 : yaml-files ( paths -- paths' ) [ yaml-file? ] filter ;
170 : uses-yaml? ( paths -- ? ) [ yaml-file? ] any? ;
172 : docker-file? ( path -- ? ) >lower file-name { "dockerfile" ".dockerignore" "docker-compose.yaml" } member? ;
173 : docker-files ( paths -- paths' ) [ docker-file? ] filter ;
174 : uses-docker? ( paths -- ? ) [ docker-file? ] any? ;
176 : automake-file? ( path -- ? )
179 [ "makefile.am" tail? ]
180 [ "makefile.am.inc" tail? ]
182 : automake-files ( paths -- paths' ) [ automake-file? ] filter ;
183 : uses-automake? ( paths -- ? ) [ automake-file? ] any? ;
185 : make-file? ( path -- ? )
186 >lower file-name { "gnumakefile" "makefile" } member? ;
187 : make-files ( paths -- paths' ) [ make-file? ] filter ;
188 : uses-make? ( paths -- ? ) [ make-file? ] any? ;
190 : nmake-file? ( path -- ? ) >lower file-name "nmakefile" = ;
191 : nmake-files ( paths -- paths' ) [ nmake-file? ] filter ;
192 : uses-nmake? ( paths -- ? ) [ nmake-file? ] any? ;
194 : gradle-file? ( path -- ? ) >lower { [ "gradle" head? ] [ ".gradle" tail? ] } 1|| ;
195 : gradle-files ( paths -- paths' ) [ gradle-file? ] filter ;
196 : uses-gradle? ( paths -- ? ) [ gradle-file? ] any? ;
198 : github-file? ( path -- ? ) >lower ".github" swap subseq? ;
199 : github-files ( paths -- paths' ) [ github-file? ] filter ;
200 : has-github-files? ( paths -- ? ) [ github-file? ] any? ;
202 : cargo-file? ( path -- ? ) file-name { "Cargo.toml" "Cargo.lock" } member? ;
203 : cargo-files ( paths -- paths' ) [ cargo-file? ] filter ;
204 : has-cargo-files? ( paths -- ? ) [ cargo-file? ] any? ;
206 : julia-project-file? ( path -- ? ) file-name { "Project.toml" } member? ;
207 : julia-project-files ( paths -- paths' ) [ julia-project-file? ] filter ;
208 : has-julia-project-files? ( paths -- ? ) [ julia-project-file? ] any? ;
210 : rust-project-dir? ( path -- ? ) file-name "Cargo.toml" = ;
212 : rust-source-file? ( path -- ? )
217 : rust-source-files ( path -- paths ) [ rust-source-file? ] filter ;
219 : rust-build-system-files ( path -- ? )
221 [ "Carg.toml" tail? ]
222 [ "Carg.lock" tail? ]
225 : rust-intermediate-build-files ( path -- ? )
232 : rust-output-files ( path -- ? )
240 : print-rust-package ( assoc -- )
242 [ "name" of [ " name: %s" sprintf print ] when* ]
243 [ "version" of [ " version: %s" sprintf print ] when* ]
244 [ "license" of [ " license: %s" sprintf print ] when* ]
245 [ "edition" of [ " rust edition: %s" sprintf print ] when* ]
248 : analyze-rust-cargo-toml ( assoc -- )
249 [ print-rust-package ] keep
250 [ "workspace" of "members" of length [ " %d member projects" sprintf print ] unless-zero ]
252 [ [ "package" of ] [ "workspace" of "package" of ] bi assoc-union ] keep
253 "workspace" of "members" of [
254 "package: " write print print-rust-package
258 : analyze-rust-project ( path -- )
259 [ "Analyzing rust project at %s" sprintf print ]
260 [ path>toml analyze-rust-cargo-toml ]
261 [ containing-directory recursive-directory-files ] tri
263 [ rust-source-files length " %d rust source files" sprintf print ]
266 : web-file? ( path -- ? )
267 >lower file-extension
269 "css" "scss" "js" "jsx" "ejs" "mjs" "ts" "tsx" "json" "html"
270 "less" "mustache" "snap" "wasm"
272 : web-files ( paths -- paths' ) [ web-file? ] filter ;
274 : rc-file? ( path -- ? ) >lower file-name { [ "." head? ] [ "rc" tail? ] } 1&& ;
275 : rc-files ( paths -- paths' ) [ rc-file? ] filter ;
277 : env-file? ( path -- ? ) >lower ".env" tail? ;
278 : env-files ( paths -- paths' ) [ env-file? ] filter ;
280 : image-file? ( path -- ? ) >lower file-extension { "png" "jpg" "jpeg" "ico" } member? ;
281 : image-files ( paths -- paths' ) [ image-file? ] filter ;
283 : ignore-file? ( path -- ? ) >lower file-name { [ "." head? ] [ "ignore" tail? ] } 1&& ;
284 : ignore-files ( paths -- paths' ) [ ignore-file? ] filter ;
286 : has-package-json? ( path -- ? ) "package.json" append-path file-exists? ;
287 : uses-git? ( path -- ? ) ".git" append-path file-exists? ;
289 : diff-paths ( paths quot -- paths' )
290 keep swap [ [ normalize-path ] map ] bi@ diff ; inline
292 : assoc. ( assoc -- )
293 [ commas ] map-values simple-table. ;
295 : analyze-codebase-path ( path -- )
297 [ normalize-path "project at path `%s`" sprintf print nl ]
298 [ uses-git? [ "uses git" print ] when ]
299 [ has-package-json? [ "has a package.json file" print ] when ]
302 : analyze-codebase-paths ( paths -- )
306 [ length "%d binary files" sprintf print ]
307 [ length "%d text files" sprintf print ] bi*
309 [ github-files [ sort "has .github files" print ... ] unless-empty ]
310 [ license-files [ sort [ length "has %d license files" sprintf print ] [ ... ] bi ] unless-empty ]
311 [ readme-files [ sort "has readme files" print ... ] unless-empty ]
312 [ owners-files [ sort "has owners files" print ... ] unless-empty ]
313 [ codenotify-files [ sort "has codenotify files" print ... ] unless-empty ]
314 [ contributing-files [ sort "has contributing files" print ... ] unless-empty ]
315 [ changelog-files [ sort "has changelog files" print ... ] unless-empty ]
316 [ security-files [ sort "has security files" print ... ] unless-empty ]
317 [ notice-files [ sort "has notice files" print ... ] unless-empty ]
318 [ version-files [ sort "has version files" print ... ] unless-empty ]
320 { [ dot-files ] [ rc-files diff ] [ ignore-files diff ] } cleave
321 [ sort "has dot files" print ... ] unless-empty
323 [ rc-files [ sort [ length "has %d rc files" sprintf print ] [ ... ] bi ] unless-empty ]
324 [ configure-files [ sort "uses configure files" print ... ] unless-empty ]
325 [ automake-files [ sort "uses automake" print ... ] unless-empty ]
326 [ make-files [ sort "uses make" print ... ] unless-empty ]
327 [ nmake-files [ sort "uses nmake" print ... ] unless-empty ]
328 [ cmake-files [ sort "uses cmake" print ... ] unless-empty ]
329 [ gradle-files [ sort "uses gradle" print ... ] unless-empty ]
330 [ cargo-files [ sort "uses rust/cargo" print ... ] unless-empty ]
331 [ julia-project-files [ sort "uses julia Project.toml" print ... ] unless-empty ]
332 [ in-files [ sort "uses 'in' files" print ... ] unless-empty ]
333 [ ignore-files [ sort [ length "has %d ignore files" sprintf print ] [ ... ] bi ] unless-empty nl ]
334 [ [ rust-project-dir? ] filter [ [ "rust projects at " print . ] [ [ analyze-rust-project ] each ] bi ] unless-empty nl ]
338 [ github-files diff ]
339 [ license-files diff ]
340 [ readme-files diff ]
341 [ owners-files diff ]
342 [ codenotify-files diff ]
343 [ contributing-files diff ]
344 [ changelog-files diff ]
345 [ security-files diff ]
346 [ notice-files diff ]
347 [ version-files diff ]
349 [ sort [ length "has %d UPPER files (minus github,license,readme,owner,codenotify,contributing,changelog,security,notice,version)" sprintf print ] [ ... ] bi ] unless-empty nl
351 [ "Top 20 largest files" print file-sizes sort-values 20 index-or-length tail* [ normalize-path ] map-keys reverse assoc. nl ]
352 [ "Top 10 file extension sizes" print sum-sizes-by-extension 10 index-or-length tail* reverse assoc. nl ]
353 [ "Top 10 text file line counts" print sum-line-counts-by-extension 10 index-or-length tail* reverse assoc. nl ]
354 [ "Top 10 file extension counts" print count-by-file-extension 10 index-or-length tail* reverse assoc. nl ]
357 : analyze-codebase ( path -- )
358 [ analyze-codebase-path ]
359 [ codebase-paths analyze-codebase-paths ] bi ;
361 : analyze-codebases ( path -- )
362 [ directory-files ] keep [ prepend-path ] curry map
363 [ file-info directory? ] filter
364 [ analyze-codebase ] each ;