@tricoteuses/senat 1.3.1 → 1.3.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (253) hide show
  1. package/lib/config.d.ts +1 -0
  2. package/lib/config.js +14 -45
  3. package/lib/databases.js +86 -143
  4. package/lib/datasets.js +78 -83
  5. package/lib/index.d.ts +7 -4
  6. package/lib/index.js +42 -419
  7. package/lib/loaders.js +149 -654
  8. package/lib/model/ameli.js +83 -21
  9. package/lib/model/debats.js +0 -1
  10. package/lib/model/dosleg.d.ts +1 -1
  11. package/lib/model/dosleg.js +179 -73
  12. package/lib/model/index.d.ts +3 -3
  13. package/lib/model/index.js +12 -46
  14. package/lib/model/questions.js +68 -39
  15. package/lib/model/sens.js +383 -113
  16. package/lib/model/texte.js +220 -290
  17. package/lib/model/util.js +9 -26
  18. package/lib/raw_types/ameli.js +5 -6
  19. package/lib/raw_types/debats.js +5 -6
  20. package/lib/raw_types/dosleg.js +5 -6
  21. package/lib/raw_types/questions.js +5 -6
  22. package/lib/raw_types/sens.js +5 -6
  23. package/lib/raw_types_schemats/ameli.js +1 -43
  24. package/lib/raw_types_schemats/debats.js +1 -22
  25. package/lib/raw_types_schemats/dosleg.js +1 -96
  26. package/lib/raw_types_schemats/questions.js +1 -22
  27. package/lib/raw_types_schemats/sens.js +1 -112
  28. package/lib/scripts/convert_data.js +181 -631
  29. package/lib/scripts/datautil.js +17 -60
  30. package/lib/scripts/parse_textes.js +46 -129
  31. package/lib/scripts/retrieve_documents.js +247 -513
  32. package/lib/scripts/retrieve_open_data.js +211 -368
  33. package/lib/scripts/retrieve_senateurs_photos.js +144 -239
  34. package/lib/scripts/shared/cli_helpers.js +30 -30
  35. package/lib/scripts/shared/util.js +28 -94
  36. package/lib/strings.js +20 -45
  37. package/lib/types/ameli.d.ts +1 -1
  38. package/lib/types/ameli.js +14 -25
  39. package/lib/types/debats.d.ts +1 -1
  40. package/lib/types/debats.js +3 -21
  41. package/lib/types/dosleg.d.ts +1 -1
  42. package/lib/types/dosleg.js +152 -119
  43. package/lib/types/questions.d.ts +1 -1
  44. package/lib/types/questions.js +1 -13
  45. package/lib/types/sens.d.ts +1 -1
  46. package/lib/types/sens.js +1 -13
  47. package/lib/types/sessions.js +44 -49
  48. package/lib/types/texte.js +17 -22
  49. package/lib/validators/config.js +47 -111
  50. package/lib/validators/senat.js +1 -5
  51. package/package.json +16 -38
  52. package/lib/aggregates.d.ts +0 -52
  53. package/lib/aggregates.mjs +0 -930
  54. package/lib/aggregates.ts +0 -833
  55. package/lib/config.mjs +0 -16
  56. package/lib/config.ts +0 -26
  57. package/lib/data/legislatures.json +0 -38
  58. package/lib/databases.mjs +0 -57
  59. package/lib/databases.ts +0 -71
  60. package/lib/datasets.mjs +0 -78
  61. package/lib/datasets.ts +0 -118
  62. package/lib/fields.d.ts +0 -10
  63. package/lib/fields.mjs +0 -68
  64. package/lib/fields.ts +0 -29
  65. package/lib/index.mjs +0 -4
  66. package/lib/index.ts +0 -42
  67. package/lib/inserters.d.ts +0 -98
  68. package/lib/inserters.mjs +0 -500
  69. package/lib/inserters.ts +0 -521
  70. package/lib/loaders.mjs +0 -158
  71. package/lib/loaders.ts +0 -271
  72. package/lib/model/ameli.mjs +0 -84
  73. package/lib/model/ameli.ts +0 -100
  74. package/lib/model/debats.mjs +0 -1
  75. package/lib/model/debats.ts +0 -0
  76. package/lib/model/dosleg.mjs +0 -196
  77. package/lib/model/dosleg.ts +0 -240
  78. package/lib/model/index.mjs +0 -4
  79. package/lib/model/index.ts +0 -14
  80. package/lib/model/questions.mjs +0 -71
  81. package/lib/model/questions.ts +0 -93
  82. package/lib/model/sens.mjs +0 -415
  83. package/lib/model/sens.ts +0 -516
  84. package/lib/model/texte.mjs +0 -208
  85. package/lib/model/texte.ts +0 -229
  86. package/lib/model/util.mjs +0 -19
  87. package/lib/model/util.ts +0 -32
  88. package/lib/raw_types/ameli.mjs +0 -5
  89. package/lib/raw_types/ameli.ts +0 -951
  90. package/lib/raw_types/debats.mjs +0 -5
  91. package/lib/raw_types/debats.ts +0 -222
  92. package/lib/raw_types/dosleg.mjs +0 -5
  93. package/lib/raw_types/dosleg.ts +0 -3625
  94. package/lib/raw_types/questions.mjs +0 -5
  95. package/lib/raw_types/questions.ts +0 -427
  96. package/lib/raw_types/sens.mjs +0 -5
  97. package/lib/raw_types/sens.ts +0 -4499
  98. package/lib/raw_types_kysely/ameli.d.ts +0 -6
  99. package/lib/raw_types_kysely/ameli.mjs +0 -7
  100. package/lib/raw_types_kysely/ameli.ts +0 -6
  101. package/lib/raw_types_kysely/debats.d.ts +0 -6
  102. package/lib/raw_types_kysely/debats.mjs +0 -7
  103. package/lib/raw_types_kysely/debats.ts +0 -6
  104. package/lib/raw_types_kysely/dosleg.d.ts +0 -6
  105. package/lib/raw_types_kysely/dosleg.mjs +0 -7
  106. package/lib/raw_types_kysely/dosleg.ts +0 -6
  107. package/lib/raw_types_kysely/questions.d.ts +0 -6
  108. package/lib/raw_types_kysely/questions.mjs +0 -7
  109. package/lib/raw_types_kysely/questions.ts +0 -6
  110. package/lib/raw_types_kysely/sens.d.ts +0 -6
  111. package/lib/raw_types_kysely/sens.mjs +0 -7
  112. package/lib/raw_types_kysely/sens.ts +0 -6
  113. package/lib/raw_types_kysely/texte.d.ts +0 -45
  114. package/lib/raw_types_kysely/texte.mjs +0 -7
  115. package/lib/raw_types_kysely/texte.ts +0 -53
  116. package/lib/raw_types_schemats/ameli.mjs +0 -2
  117. package/lib/raw_types_schemats/ameli.ts +0 -601
  118. package/lib/raw_types_schemats/debats.mjs +0 -2
  119. package/lib/raw_types_schemats/debats.ts +0 -145
  120. package/lib/raw_types_schemats/dosleg.mjs +0 -2
  121. package/lib/raw_types_schemats/dosleg.ts +0 -2195
  122. package/lib/raw_types_schemats/questions.mjs +0 -2
  123. package/lib/raw_types_schemats/questions.ts +0 -251
  124. package/lib/raw_types_schemats/sens.mjs +0 -2
  125. package/lib/raw_types_schemats/sens.ts +0 -2907
  126. package/lib/scripts/convert_data.mjs +0 -181
  127. package/lib/scripts/convert_data.ts +0 -243
  128. package/lib/scripts/datautil.mjs +0 -16
  129. package/lib/scripts/datautil.ts +0 -19
  130. package/lib/scripts/images/transparent_150x192.jpg +0 -0
  131. package/lib/scripts/images/transparent_155x225.jpg +0 -0
  132. package/lib/scripts/parse_textes.mjs +0 -46
  133. package/lib/scripts/parse_textes.ts +0 -65
  134. package/lib/scripts/retrieve_documents.mjs +0 -249
  135. package/lib/scripts/retrieve_documents.ts +0 -298
  136. package/lib/scripts/retrieve_open_data.mjs +0 -217
  137. package/lib/scripts/retrieve_open_data.ts +0 -274
  138. package/lib/scripts/retrieve_senateurs_photos.mjs +0 -147
  139. package/lib/scripts/retrieve_senateurs_photos.ts +0 -177
  140. package/lib/scripts/retrieve_textes.d.ts +0 -1
  141. package/lib/scripts/retrieve_textes.mjs +0 -328
  142. package/lib/scripts/retrieve_textes.ts +0 -143
  143. package/lib/scripts/shared/cli_helpers.ts +0 -36
  144. package/lib/scripts/shared/util.ts +0 -33
  145. package/lib/src/aggregates.d.ts +0 -52
  146. package/lib/src/aggregates.mjs +0 -726
  147. package/lib/src/config.d.ts +0 -2
  148. package/lib/src/config.mjs +0 -16
  149. package/lib/src/databases.d.ts +0 -18
  150. package/lib/src/databases.mjs +0 -55
  151. package/lib/src/datasets.d.ts +0 -28
  152. package/lib/src/datasets.mjs +0 -78
  153. package/lib/src/fields.d.ts +0 -10
  154. package/lib/src/fields.mjs +0 -22
  155. package/lib/src/index.d.ts +0 -8
  156. package/lib/src/index.mjs +0 -7
  157. package/lib/src/inserters.d.ts +0 -98
  158. package/lib/src/inserters.mjs +0 -360
  159. package/lib/src/loaders.d.ts +0 -36
  160. package/lib/src/loaders.mjs +0 -107
  161. package/lib/src/model/ameli.d.ts +0 -4
  162. package/lib/src/model/ameli.js +0 -57
  163. package/lib/src/model/debats.d.ts +0 -4
  164. package/lib/src/model/debats.js +0 -43
  165. package/lib/src/model/dosleg.d.ts +0 -197
  166. package/lib/src/model/dosleg.js +0 -169
  167. package/lib/src/model/index.d.ts +0 -4
  168. package/lib/src/model/index.js +0 -4
  169. package/lib/src/model/questions.d.ts +0 -89
  170. package/lib/src/model/questions.js +0 -76
  171. package/lib/src/model/sens.d.ts +0 -390
  172. package/lib/src/model/sens.js +0 -339
  173. package/lib/src/model/texte.d.ts +0 -7
  174. package/lib/src/model/texte.js +0 -183
  175. package/lib/src/raw_types_kysely/ameli.d.ts +0 -915
  176. package/lib/src/raw_types_kysely/ameli.js +0 -5
  177. package/lib/src/raw_types_kysely/debats.d.ts +0 -207
  178. package/lib/src/raw_types_kysely/debats.js +0 -5
  179. package/lib/src/raw_types_kysely/dosleg.d.ts +0 -3532
  180. package/lib/src/raw_types_kysely/dosleg.js +0 -5
  181. package/lib/src/raw_types_kysely/questions.d.ts +0 -414
  182. package/lib/src/raw_types_kysely/questions.js +0 -5
  183. package/lib/src/raw_types_kysely/sens.d.ts +0 -4394
  184. package/lib/src/raw_types_kysely/sens.js +0 -5
  185. package/lib/src/raw_types_schemats/ameli.d.ts +0 -541
  186. package/lib/src/raw_types_schemats/ameli.js +0 -2
  187. package/lib/src/raw_types_schemats/debats.d.ts +0 -127
  188. package/lib/src/raw_types_schemats/debats.js +0 -2
  189. package/lib/src/raw_types_schemats/dosleg.d.ts +0 -2027
  190. package/lib/src/raw_types_schemats/dosleg.js +0 -2
  191. package/lib/src/raw_types_schemats/questions.d.ts +0 -231
  192. package/lib/src/raw_types_schemats/questions.js +0 -2
  193. package/lib/src/raw_types_schemats/sens.d.ts +0 -2709
  194. package/lib/src/raw_types_schemats/sens.js +0 -2
  195. package/lib/src/scripts/convert_data.d.ts +0 -1
  196. package/lib/src/scripts/convert_data.js +0 -95
  197. package/lib/src/scripts/datautil.d.ts +0 -5
  198. package/lib/src/scripts/datautil.js +0 -16
  199. package/lib/src/scripts/parse_textes.d.ts +0 -1
  200. package/lib/src/scripts/parse_textes.js +0 -47
  201. package/lib/src/scripts/retrieve_documents.d.ts +0 -1
  202. package/lib/src/scripts/retrieve_documents.js +0 -258
  203. package/lib/src/scripts/retrieve_open_data.d.ts +0 -1
  204. package/lib/src/scripts/retrieve_open_data.js +0 -214
  205. package/lib/src/scripts/retrieve_senateurs_photos.d.ts +0 -1
  206. package/lib/src/scripts/retrieve_senateurs_photos.js +0 -147
  207. package/lib/src/scripts/shared/cli_helpers.d.ts +0 -44
  208. package/lib/src/scripts/shared/cli_helpers.js +0 -32
  209. package/lib/src/scripts/shared/util.d.ts +0 -3
  210. package/lib/src/scripts/shared/util.js +0 -28
  211. package/lib/src/strings.d.ts +0 -1
  212. package/lib/src/strings.mjs +0 -18
  213. package/lib/src/types/ameli.d.ts +0 -10
  214. package/lib/src/types/ameli.js +0 -13
  215. package/lib/src/types/debats.d.ts +0 -4
  216. package/lib/src/types/debats.js +0 -2
  217. package/lib/src/types/dosleg.d.ts +0 -98
  218. package/lib/src/types/dosleg.js +0 -151
  219. package/lib/src/types/questions.d.ts +0 -2
  220. package/lib/src/types/questions.js +0 -1
  221. package/lib/src/types/sens.d.ts +0 -10
  222. package/lib/src/types/sens.js +0 -1
  223. package/lib/src/types/sessions.d.ts +0 -42
  224. package/lib/src/types/sessions.js +0 -43
  225. package/lib/src/types/texte.d.ts +0 -61
  226. package/lib/src/types/texte.js +0 -16
  227. package/lib/src/validators/config.d.ts +0 -1
  228. package/lib/src/validators/config.js +0 -54
  229. package/lib/src/validators/senat.d.ts +0 -0
  230. package/lib/src/validators/senat.js +0 -24
  231. package/lib/strings.mjs +0 -18
  232. package/lib/strings.ts +0 -26
  233. package/lib/types/ameli.mjs +0 -13
  234. package/lib/types/ameli.ts +0 -21
  235. package/lib/types/debats.mjs +0 -2
  236. package/lib/types/debats.ts +0 -6
  237. package/lib/types/dosleg.mjs +0 -151
  238. package/lib/types/dosleg.ts +0 -284
  239. package/lib/types/questions.mjs +0 -1
  240. package/lib/types/questions.ts +0 -3
  241. package/lib/types/sens.mjs +0 -1
  242. package/lib/types/sens.ts +0 -12
  243. package/lib/types/sessions.mjs +0 -43
  244. package/lib/types/sessions.ts +0 -42
  245. package/lib/types/texte.mjs +0 -16
  246. package/lib/types/texte.ts +0 -76
  247. package/lib/typings/windows-1252.d.js +0 -2
  248. package/lib/typings/windows-1252.d.mjs +0 -2
  249. package/lib/typings/windows-1252.d.ts +0 -11
  250. package/lib/validators/config.mjs +0 -54
  251. package/lib/validators/config.ts +0 -79
  252. package/lib/validators/senat.mjs +0 -24
  253. package/lib/validators/senat.ts +0 -26
@@ -1,2 +0,0 @@
1
- /* tslint:disable */
2
- export {};
@@ -1 +0,0 @@
1
- export {};
@@ -1,95 +0,0 @@
1
- import assert from "assert";
2
- import commandLineArgs from "command-line-args";
3
- import fs from "fs-extra";
4
- import path from "path";
5
- import { datasets, EnabledDatasets, getEnabledDatasets } from "../datasets";
6
- import { findAllCirconscriptions, findAllLois, findAllQuestions, findAllSens } from "../model";
7
- import { formatToFourDigitSession, SIGNET_STRUCTURE_REGEXP, UNDEFINED_SESSION } from "./datautil";
8
- import { commonOptions } from "./shared/cli_helpers";
9
- import { ensureAndClearDir } from "./shared/util";
10
- const optionsDefinitions = [
11
- ...commonOptions,
12
- ];
13
- const options = commandLineArgs(optionsDefinitions);
14
- async function convertData() {
15
- const enabledDatasets = getEnabledDatasets(options.categories);
16
- const dataDir = options.dataDir;
17
- assert(dataDir, "Missing argument: data directory");
18
- console.time("conversion time");
19
- if (enabledDatasets & EnabledDatasets.DosLeg) {
20
- const dataset = datasets.dosleg;
21
- if (!options.silent) {
22
- console.log(`Converting database ${dataset.database} data into files…`);
23
- }
24
- const loiReorganizedRootDir = path.join(dataDir, dataset.database);
25
- ensureAndClearDir(loiReorganizedRootDir);
26
- for await (const loi of findAllLois()) {
27
- if (options.verbose) {
28
- console.log(`Converting ${loi.signet} file…`);
29
- }
30
- let loiReorganizedDir = path.join(loiReorganizedRootDir, UNDEFINED_SESSION);
31
- const signetParts = SIGNET_STRUCTURE_REGEXP.exec(loi.signet)?.groups;
32
- if (signetParts) {
33
- const { session } = signetParts;
34
- const formattedSession = formatToFourDigitSession(session);
35
- loiReorganizedDir = path.join(loiReorganizedRootDir, formattedSession);
36
- }
37
- fs.ensureDirSync(loiReorganizedDir);
38
- const loiFileName = `${loi.signet}.json`;
39
- fs.writeJSONSync(path.join(loiReorganizedDir, loiFileName), loi, { spaces: 2 });
40
- }
41
- }
42
- if (enabledDatasets & EnabledDatasets.Questions) {
43
- const dataset = datasets.questions;
44
- if (!options.silent) {
45
- console.log(`Converting database ${dataset.database} data into files…`);
46
- }
47
- const questionReorganizedRootDir = path.join(dataDir, dataset.database);
48
- ensureAndClearDir(questionReorganizedRootDir);
49
- for await (const question of findAllQuestions()) {
50
- if (options.verbose) {
51
- console.log(`Converting ${question.reference} file…`);
52
- }
53
- const legislature = question.legislature ? question.legislature : 0;
54
- const questionReorganizedDir = path.join(questionReorganizedRootDir, String(legislature));
55
- fs.ensureDirSync(questionReorganizedDir);
56
- const questionFileName = `${question.reference}.json`;
57
- fs.writeJSONSync(path.join(questionReorganizedDir, questionFileName), question, { spaces: 2 });
58
- }
59
- }
60
- if (enabledDatasets & EnabledDatasets.Sens) {
61
- const dataset = datasets.sens;
62
- if (!options.silent) {
63
- console.log(`Converting database ${dataset.database} data into files…`);
64
- }
65
- const senReorganizedRootDir = path.join(dataDir, dataset.database);
66
- const senReorganizedDir = path.join(senReorganizedRootDir, "sens");
67
- const circonscriptionsReorganizedDir = path.join(senReorganizedRootDir, "circonscriptions");
68
- ensureAndClearDir(senReorganizedRootDir);
69
- ensureAndClearDir(senReorganizedDir);
70
- ensureAndClearDir(circonscriptionsReorganizedDir);
71
- for await (const sen of findAllSens()) {
72
- if (options.verbose) {
73
- console.log(`Converting ${sen.matricule} file…`);
74
- }
75
- const senFileName = `${sen.matricule}.json`;
76
- fs.writeJSONSync(path.join(senReorganizedDir, senFileName), sen, { spaces: 2 });
77
- }
78
- for await (const circonscription of findAllCirconscriptions()) {
79
- if (options.verbose) {
80
- console.log(`Converting ${circonscription.identifiant} file…`);
81
- }
82
- const circonscriptionFileName = `${circonscription.identifiant}.json`;
83
- fs.writeJSONSync(path.join(circonscriptionsReorganizedDir, circonscriptionFileName), circonscription, { spaces: 2 });
84
- }
85
- }
86
- if (!options.silent) {
87
- console.timeEnd("conversion time");
88
- }
89
- }
90
- convertData()
91
- .then(() => process.exit(0))
92
- .catch((error) => {
93
- console.log(error);
94
- process.exit(1);
95
- });
@@ -1,5 +0,0 @@
1
- export declare const SIGNET_STRUCTURE_REGEXP: RegExp;
2
- export declare const AKN_IDENTIFICATION_STRUCTURE_REGEXP: RegExp;
3
- export declare const AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP: RegExp;
4
- export declare const UNDEFINED_SESSION = "0";
5
- export declare function formatToFourDigitSession(session: string): string;
@@ -1,16 +0,0 @@
1
- import { DateTime, Settings } from "luxon";
2
- Settings.twoDigitCutoffYear = 50;
3
- export const SIGNET_STRUCTURE_REGEXP = /^(?<type>[a-z]+)(?<session>\d{2,4})-?(?<numTexte>\d+)?/;
4
- export const AKN_IDENTIFICATION_STRUCTURE_REGEXP = /^\/akn\/fr\/(?<type>[a-z]+)\/(?<session>\d{4}-\d{4})\/?(?<numTexte>\d+)\/fr@(?<version>\b(?:RECT|RECT_BIS|RECT_TER|RECT_QUATER|RECT_QUINQUIES)\b)?/;
5
- export const AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP = /^\/akn\/fr\/(?<type>[a-z]+)\/(?<session>\d{2,4})\/?(?<numTexte>[a-zA-Z0-9]+)\/fr@(?<version>\b(?:RECT|RECT_BIS|RECT_TER|RECT_QUATER|RECT_QUINQUIES)\b)?/;
6
- export const UNDEFINED_SESSION = "0";
7
- export function formatToFourDigitSession(session) {
8
- if (session.length >= 2) {
9
- const sessionFirstTwoDigits = session.substring(0, 2);
10
- const sessionLastTwoDigits = session.substring(session.length - 2);
11
- const twoDigitSession = parseInt(sessionFirstTwoDigits) === parseInt(sessionLastTwoDigits) - 1 ?
12
- sessionFirstTwoDigits : sessionLastTwoDigits;
13
- return DateTime.fromFormat(String(twoDigitSession), "yy").toFormat("yyyy");
14
- }
15
- return UNDEFINED_SESSION;
16
- }
@@ -1 +0,0 @@
1
- export {};
@@ -1,47 +0,0 @@
1
- import assert from "assert";
2
- import commandLineArgs from "command-line-args";
3
- import fs from "fs-extra";
4
- import path from "path";
5
- import { iterFilePaths } from "../loaders";
6
- import { parseExposeDesMotifsFromFile, parseTexteFromFile } from "../model/texte";
7
- import { commonOptions } from "./shared/cli_helpers";
8
- import { ensureAndClearDir } from "./shared/util";
9
- const optionsDefinitions = [
10
- ...commonOptions,
11
- ];
12
- const options = commandLineArgs(optionsDefinitions);
13
- async function main() {
14
- const dataDir = options.dataDir;
15
- assert(dataDir, "Missing argument: data directory");
16
- const transformedTextesDir = path.join(options.dataDir, "leg", "transformed");
17
- ensureAndClearDir(transformedTextesDir);
18
- for (const filePath of iterFilePaths(path.join(dataDir, "leg", "original"))) {
19
- const parsedFilePath = path.parse(filePath);
20
- if (parsedFilePath.ext !== ".xml") {
21
- continue;
22
- }
23
- const texteDirFromOriginal = parsedFilePath.dir.substring(filePath.indexOf("original") + "original".length);
24
- const transformedTexteDir = path.join(transformedTextesDir, texteDirFromOriginal);
25
- fs.ensureDirSync(transformedTexteDir);
26
- if (!options.silent) {
27
- console.log(`Parsing texte ${parsedFilePath.name}.xml…`);
28
- }
29
- const parsedFile = await parseTexteFromFile(filePath);
30
- const exposeDesMotifsFileName = `${parsedFilePath.name
31
- .substring(0, parsedFilePath.name.indexOf(".akn"))}-expose`;
32
- const exposeDesMotifsFilePath = path.join(parsedFilePath.dir, `${exposeDesMotifsFileName}.html`);
33
- if (parsedFile && fs.existsSync(exposeDesMotifsFilePath)) {
34
- if (!options.silent) {
35
- console.log(`Parsing exposé des motifs ${exposeDesMotifsFileName}.html…`);
36
- }
37
- parsedFile.exposeDesMotifs = await parseExposeDesMotifsFromFile(exposeDesMotifsFilePath);
38
- }
39
- fs.writeJSONSync(path.join(transformedTexteDir, `${parsedFilePath.name}.json`), parsedFile, { spaces: 2 });
40
- }
41
- }
42
- main()
43
- .then(() => process.exit(0))
44
- .catch((error) => {
45
- console.log(error);
46
- process.exit(1);
47
- });
@@ -1 +0,0 @@
1
- export {};
@@ -1,258 +0,0 @@
1
- import assert from "assert";
2
- import commandLineArgs from "command-line-args";
3
- import fs from "fs-extra";
4
- import path from "path";
5
- import { findSenatRapportUrls, findSenatTexteUrls } from "../model/dosleg";
6
- import { parseTexte, parseTexteFromFile } from "../model/texte";
7
- import { UNDEFINED_SESSION } from "./datautil";
8
- import { commonOptions } from "./shared/cli_helpers";
9
- import { ensureAndClearDir, fetchWithRetry, isOptionEmptyOrHasValue } from "./shared/util";
10
- const optionsDefinitions = [
11
- ...commonOptions,
12
- {
13
- help: "sessions of textes to retrieve; leave empty for all",
14
- multiple: true,
15
- name: "sessions",
16
- type: String,
17
- },
18
- {
19
- help: "parse and convert documents into JSON (textes only for now, requires format xml)",
20
- name: "parseDocuments",
21
- type: Boolean,
22
- },
23
- {
24
- alias: "F",
25
- help: "formats of documents to retrieve (xml/html/pdf for textes, html/pdf for rapports); leave empty for all",
26
- multiple: true,
27
- name: "formats",
28
- type: String,
29
- },
30
- {
31
- help: "types of documents to retrieve (textes/rapports); leave empty for all",
32
- multiple: true,
33
- name: "types",
34
- type: String,
35
- },
36
- {
37
- help: "force retrieve all documents, even already retrieved ones",
38
- name: "force",
39
- type: Boolean,
40
- },
41
- ];
42
- const options = commandLineArgs(optionsDefinitions);
43
- const SENAT_TEXTE_XML_BASE_URL = "https://www.senat.fr/akomantoso/";
44
- const SENAT_TEXTE_BASE_URL = "https://www.senat.fr/leg/";
45
- const SENAT_EXPOSE_DES_MOTIFS_BASE_URL = "https://www.senat.fr/leg/exposes-des-motifs/";
46
- const SENAT_RAPPORT_BASE_URL = "https://www.senat.fr/rap/";
47
- const textDecoder = new TextDecoder("utf8");
48
- async function retrieveDocument(documentUrl) {
49
- if (!options.silent) {
50
- console.log(`Retrieving document ${documentUrl}…`);
51
- }
52
- try {
53
- const response = await fetchWithRetry(documentUrl);
54
- if (!response.ok) {
55
- if (response.status === 404) {
56
- console.warn(`Texte ${documentUrl} not found`);
57
- }
58
- else {
59
- console.error(`An error occurred while retrieving texte ${documentUrl}: ${response.status}`);
60
- }
61
- return null;
62
- }
63
- return response.arrayBuffer();
64
- }
65
- catch (error) {
66
- console.error(error.message);
67
- return null;
68
- }
69
- }
70
- async function retrieveTextes(dataDir) {
71
- const textesDir = path.join(dataDir, "leg");
72
- fs.ensureDirSync(textesDir);
73
- const originalTextesDir = path.join(textesDir, "original");
74
- const transformedTextesDir = path.join(textesDir, "transformed");
75
- ensureAndClearDir(transformedTextesDir);
76
- let retrievedTextesCount = 0;
77
- const texteUrlsNotFoundOrError = [];
78
- const texteUrlsParseError = [];
79
- for await (const texte of findSenatTexteUrls(options.sessions)) {
80
- const texteName = path.parse(texte.url).name;
81
- const texteDir = path.join(originalTextesDir, `${texte.session ?? UNDEFINED_SESSION}`, texteName);
82
- fs.ensureDirSync(texteDir);
83
- if (isOptionEmptyOrHasValue(options.formats, "xml")) {
84
- const texteXmlUrl = `${texteName}.akn.xml`;
85
- const texteXmlAbsoluteUrl = new URL(texteXmlUrl, SENAT_TEXTE_XML_BASE_URL).toString();
86
- const textePath = path.join(texteDir, texteXmlUrl);
87
- let texteBuffer = null;
88
- if (!options.force && fs.existsSync(textePath)) {
89
- if (!options.silent) {
90
- console.info(`Already retrieved texte ${textePath}…`);
91
- }
92
- }
93
- else {
94
- texteBuffer = await retrieveDocument(texteXmlAbsoluteUrl);
95
- if (!texteBuffer) {
96
- texteUrlsNotFoundOrError.push(texteXmlAbsoluteUrl);
97
- continue;
98
- }
99
- fs.writeFileSync(textePath, Buffer.from(texteBuffer));
100
- retrievedTextesCount++;
101
- }
102
- if (options.parseDocuments) {
103
- if (!options.silent) {
104
- console.log(`Parsing texte ${texteXmlUrl}…`);
105
- }
106
- let parsedTexte = null;
107
- if (texteBuffer) {
108
- const texteXml = textDecoder.decode(texteBuffer);
109
- parsedTexte = parseTexte(texteXml);
110
- }
111
- else {
112
- parsedTexte = await parseTexteFromFile(textePath);
113
- }
114
- if (!parsedTexte) {
115
- texteUrlsParseError.push(texteXmlAbsoluteUrl);
116
- continue;
117
- }
118
- const transformedTexteDir = path.join(transformedTextesDir, `${texte.session ?? UNDEFINED_SESSION}`, texteName);
119
- fs.ensureDirSync(transformedTexteDir);
120
- fs.writeJSONSync(path.join(transformedTexteDir, `${texteName}.akn.json`), parsedTexte, { spaces: 2 });
121
- }
122
- }
123
- if (isOptionEmptyOrHasValue(options.formats, "html")) {
124
- const texteHtmlUrl = `${texteName}.html`;
125
- const texteHtmlAbsoluteUrl = new URL(texteHtmlUrl, SENAT_TEXTE_BASE_URL).toString();
126
- const textePath = path.join(texteDir, texteHtmlUrl);
127
- if (!options.force && fs.existsSync(textePath)) {
128
- if (!options.silent) {
129
- console.info(`Already retrieved texte ${textePath}…`);
130
- }
131
- }
132
- else {
133
- const texteBuffer = await retrieveDocument(texteHtmlAbsoluteUrl);
134
- if (!texteBuffer) {
135
- texteUrlsNotFoundOrError.push(texteHtmlAbsoluteUrl);
136
- continue;
137
- }
138
- fs.writeFileSync(textePath, Buffer.from(texteBuffer));
139
- retrievedTextesCount++;
140
- }
141
- }
142
- if (isOptionEmptyOrHasValue(options.formats, "pdf")) {
143
- const textePdfUrl = `${texteName}.pdf`;
144
- const textePdfAbsoluteUrl = new URL(textePdfUrl, SENAT_TEXTE_BASE_URL).toString();
145
- const textePath = path.join(texteDir, textePdfUrl);
146
- if (!options.force && fs.existsSync(textePath)) {
147
- if (!options.silent) {
148
- console.info(`Already retrieved texte ${textePath}…`);
149
- }
150
- }
151
- else {
152
- const texteBuffer = await retrieveDocument(textePdfAbsoluteUrl);
153
- if (!texteBuffer) {
154
- texteUrlsNotFoundOrError.push(textePdfAbsoluteUrl);
155
- continue;
156
- }
157
- fs.writeFileSync(textePath, Buffer.from(texteBuffer));
158
- retrievedTextesCount++;
159
- }
160
- }
161
- if (texte.hasExposeDesMotifs) {
162
- if (options.verbose) {
163
- console.log("Retrieving exposé des motifs…");
164
- }
165
- const exposeDesMotifsUrl = `${texteName}-expose.html`;
166
- const exposeDesMotifsAbsoluteUrl = new URL(exposeDesMotifsUrl, SENAT_EXPOSE_DES_MOTIFS_BASE_URL).toString();
167
- const exposeDesMotifsPath = path.join(texteDir, exposeDesMotifsUrl);
168
- const texteBuffer = await retrieveDocument(exposeDesMotifsAbsoluteUrl);
169
- if (!texteBuffer) {
170
- continue;
171
- }
172
- fs.writeFileSync(exposeDesMotifsPath, Buffer.from(texteBuffer));
173
- }
174
- }
175
- if (options.verbose) {
176
- console.log(`${retrievedTextesCount} textes retrieved`);
177
- console.log(`${texteUrlsNotFoundOrError.length} textes failed to be retrieved with URLs ${texteUrlsNotFoundOrError.join(", ")}`);
178
- if (options.parseDocuments) {
179
- console.log(`${texteUrlsParseError.length} textes failed to be parsed with URLs ${texteUrlsParseError.join(", ")}`);
180
- }
181
- }
182
- }
183
- async function retrieveRapports(dataDir) {
184
- const rapportsDir = path.join(dataDir, "rap");
185
- fs.ensureDirSync(rapportsDir);
186
- let retrievedRapportsCount = 0;
187
- const rapportUrlsNotFoundOrError = [];
188
- for await (const rapport of findSenatRapportUrls(options.sessions)) {
189
- const parsedRapportUrl = path.parse(rapport.url);
190
- const rapportName = parsedRapportUrl.name;
191
- const rapportDir = path.join(rapportsDir, `${rapport.session ?? UNDEFINED_SESSION}`, rapportName);
192
- fs.ensureDirSync(rapportDir);
193
- if (isOptionEmptyOrHasValue(options.formats, "html")) {
194
- const rapportHtmlUrlBase = `${rapportName}_mono.html`;
195
- const rapportHtmlUrl = path.format({
196
- dir: parsedRapportUrl.dir,
197
- base: rapportHtmlUrlBase,
198
- });
199
- const rapportHtmlAbsoluteUrl = new URL(rapportHtmlUrl, SENAT_RAPPORT_BASE_URL).toString();
200
- const rapportPath = path.join(rapportDir, rapportHtmlUrlBase);
201
- if (!options.force && fs.existsSync(rapportPath)) {
202
- if (!options.silent) {
203
- console.info(`Already retrieved rapport ${rapportPath}…`);
204
- }
205
- continue;
206
- }
207
- const rapportBuffer = await retrieveDocument(rapportHtmlAbsoluteUrl);
208
- if (!rapportBuffer) {
209
- rapportUrlsNotFoundOrError.push(rapportHtmlAbsoluteUrl);
210
- continue;
211
- }
212
- fs.writeFileSync(rapportPath, Buffer.from(rapportBuffer));
213
- retrievedRapportsCount++;
214
- }
215
- if (isOptionEmptyOrHasValue(options.formats, "pdf")) {
216
- const rapportPdfUrlBase = `${rapportName}1.pdf`;
217
- const rapportPdfUrl = path.format({
218
- dir: parsedRapportUrl.dir,
219
- base: rapportPdfUrlBase,
220
- });
221
- const rapportPdfAbsoluteUrl = new URL(rapportPdfUrl, SENAT_RAPPORT_BASE_URL).toString();
222
- const rapportPath = path.join(rapportDir, rapportPdfUrlBase);
223
- if (!options.force && fs.existsSync(rapportPath)) {
224
- if (!options.silent) {
225
- console.info(`Already retrieved rapport ${rapportPath}…`);
226
- }
227
- continue;
228
- }
229
- const rapportBuffer = await retrieveDocument(rapportPdfAbsoluteUrl);
230
- if (!rapportBuffer) {
231
- rapportUrlsNotFoundOrError.push(rapportPdfAbsoluteUrl);
232
- continue;
233
- }
234
- fs.writeFileSync(rapportPath, Buffer.from(rapportBuffer));
235
- retrievedRapportsCount++;
236
- }
237
- }
238
- if (options.verbose) {
239
- console.log(`${retrievedRapportsCount} rapports retrieved`);
240
- console.log(`${rapportUrlsNotFoundOrError.length} rapports failed with URLs ${rapportUrlsNotFoundOrError.join(", ")}`);
241
- }
242
- }
243
- async function main() {
244
- const dataDir = options.dataDir;
245
- assert(dataDir, "Missing argument: data directory");
246
- if (isOptionEmptyOrHasValue(options.types, "textes")) {
247
- await retrieveTextes(dataDir);
248
- }
249
- if (isOptionEmptyOrHasValue(options.types, "rapports")) {
250
- await retrieveRapports(dataDir);
251
- }
252
- }
253
- main()
254
- .then(() => process.exit(0))
255
- .catch((error) => {
256
- console.log(error);
257
- process.exit(1);
258
- });
@@ -1 +0,0 @@
1
- export {};
@@ -1,214 +0,0 @@
1
- import assert from "assert";
2
- import { execSync } from "child_process";
3
- import commandLineArgs from "command-line-args";
4
- import fs from "fs-extra";
5
- // import fetch from "node-fetch"
6
- import path from "path";
7
- // import stream from "stream"
8
- import StreamZip from "node-stream-zip";
9
- import readline from "readline";
10
- // import util from "util"
11
- import windows1252 from "windows-1252";
12
- import config from "../config";
13
- import { getChosenFromEnabledDatasets, } from "../datasets";
14
- import { commonOptions } from "./shared/cli_helpers";
15
- const badWindows1252CharacterRegex = /[\u0080-\u009f]/g;
16
- const optionsDefinitions = [
17
- ...commonOptions,
18
- {
19
- alias: "a",
20
- help: "all options: fetch, unzip, repair-encoding, import",
21
- name: "all",
22
- type: Boolean,
23
- },
24
- {
25
- alias: "f",
26
- help: "fetch datasets instead of retrieving them from files",
27
- name: "fetch",
28
- type: Boolean,
29
- },
30
- {
31
- alias: "z",
32
- help: "unzip SQL files",
33
- name: "unzip",
34
- type: Boolean,
35
- },
36
- {
37
- alias: "e",
38
- help: "repair Windows CP 1252 encoding of SQL dumps",
39
- name: "repairEncoding",
40
- type: Boolean,
41
- },
42
- {
43
- alias: "i",
44
- help: "import SQL dumps into a freshly (re-)created database",
45
- name: "import",
46
- type: Boolean,
47
- },
48
- {
49
- alias: "d",
50
- help: "repair database (update schema and types)",
51
- name: "repairDatabase",
52
- type: Boolean,
53
- },
54
- {
55
- alias: "c",
56
- help: "create TypeScript interfaces from databases schemas into src/raw_types_* directories",
57
- name: "schema",
58
- type: Boolean,
59
- },
60
- ];
61
- const options = commandLineArgs(optionsDefinitions);
62
- // const pipeline = util.promisify(stream.pipeline)
63
- async function retrieveDataset(dataDir, dataset) {
64
- const zipFilename = dataset.url.substring(dataset.url.lastIndexOf("/") + 1);
65
- const zipFilePath = path.join(dataDir, zipFilename);
66
- if (options.all || options.fetch) {
67
- // Fetch & save ZIP file.
68
- if (!options.silent) {
69
- console.log(`Loading ${dataset.title}: ${zipFilename}…`);
70
- }
71
- // Fetch fails with OpenSSL error: dh key too small.
72
- // (so does "curl").
73
- // const response = await fetch(dataset.url)
74
- // if (!response.ok) {
75
- // console.error(response.status, response.statusText)
76
- // console.error(await response.text())
77
- // throw new Error(`Fetch failed: ${dataset.url}`)
78
- // }
79
- // await pipeline(response.body!, fs.createWriteStream(zipFilePath))
80
- fs.removeSync(zipFilePath);
81
- execSync(`wget --quiet ${dataset.url}`, {
82
- cwd: dataDir,
83
- env: process.env,
84
- encoding: "utf-8",
85
- // stdio: ["ignore", "ignore", "pipe"],
86
- });
87
- }
88
- const sqlFilename = `${dataset.database}.sql`;
89
- const sqlFilePath = path.join(dataDir, sqlFilename);
90
- if (options.all || options.unzip) {
91
- if (!options.silent) {
92
- console.log(`Unzipping ${dataset.title}: ${zipFilename}…`);
93
- }
94
- fs.removeSync(sqlFilePath);
95
- const zip = new StreamZip({
96
- file: zipFilePath,
97
- storeEntries: true,
98
- });
99
- await new Promise((resolve, reject) => {
100
- zip.on("ready", () => {
101
- zip.extract(null, dataDir, (err, _count) => {
102
- zip.close();
103
- if (err) {
104
- reject(err);
105
- }
106
- else {
107
- resolve(null);
108
- }
109
- });
110
- });
111
- });
112
- if (dataset.repairZip !== undefined) {
113
- if (!options.silent) {
114
- console.log(`Repairing Zip path ${dataset.title}: ${sqlFilename}…`);
115
- }
116
- dataset.repairZip(dataset, dataDir);
117
- }
118
- }
119
- if ((options.all || options.repairEncoding) && dataset.repairEncoding) {
120
- if (!options.silent) {
121
- console.log(`Repairing Windows CP1252 encoding of ${dataset.title}: ${sqlFilename}…`);
122
- }
123
- const repairedSqlFilePath = sqlFilePath + ".repaired";
124
- const repairedSqlWriter = fs.createWriteStream(repairedSqlFilePath, {
125
- encoding: "utf8",
126
- });
127
- const lineReader = readline.createInterface({
128
- input: fs.createReadStream(sqlFilePath, { encoding: "utf8" }),
129
- crlfDelay: Infinity,
130
- });
131
- for await (const line of lineReader) {
132
- repairedSqlWriter.write(line.replace(badWindows1252CharacterRegex, (match) => windows1252.decode(match, { mode: "fatal" })) + "\n");
133
- }
134
- repairedSqlWriter.end();
135
- await fs.move(repairedSqlFilePath, sqlFilePath, { overwrite: true });
136
- }
137
- if (options.all || options.import) {
138
- if (!options.silent) {
139
- console.log(`Importing ${dataset.title}: ${sqlFilename}…`);
140
- }
141
- execSync(`psql -c "DROP DATABASE IF EXISTS ${dataset.database}"`, {
142
- cwd: dataDir,
143
- env: process.env,
144
- encoding: "utf-8",
145
- // stdio: ["ignore", "ignore", "pipe"],
146
- });
147
- execSync(`psql -c "CREATE DATABASE ${dataset.database} WITH OWNER opendata"`, {
148
- cwd: dataDir,
149
- env: process.env,
150
- encoding: "utf-8",
151
- // stdio: ["ignore", "ignore", "pipe"],
152
- });
153
- execSync(`psql -f ${sqlFilename} ${dataset.database}`, {
154
- cwd: dataDir,
155
- env: process.env,
156
- encoding: "utf-8",
157
- // stdio: ["ignore", "ignore", "pipe"],
158
- });
159
- }
160
- if (options.schema) {
161
- let definitionsDir = path.resolve("src", "raw_types_schemats");
162
- assert(fs.statSync(definitionsDir).isDirectory());
163
- if (!options.silent) {
164
- console.log(`Creating TypeScript definitions from schema of database ${dataset.database}…`);
165
- }
166
- const dbConnectionString = `postgres://${process.env.PGUSER}:${process.env.PGPASSWORD}@${process.env.PGHOST}:${process.env.PGPORT}/${dataset.database}`;
167
- let definitionFilePath = path.join(definitionsDir, `${dataset.database}.ts`);
168
- execSync(`npx schemats generate -c ${dbConnectionString} -s ${dataset.schema} -o ${definitionFilePath}`, {
169
- // cwd: dataDir,
170
- env: process.env,
171
- encoding: "utf-8",
172
- // stdio: ["ignore", "ignore", "pipe"],
173
- });
174
- const definition = fs.readFileSync(definitionFilePath, { encoding: "utf8" });
175
- const definitionRepaired = definition
176
- .replace(/\r\n/g, "\n")
177
- .replace(/AUTO-GENERATED FILE @ \d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}/, "AUTO-GENERATED FILE");
178
- fs.writeFileSync(definitionFilePath, definitionRepaired);
179
- definitionsDir = path.resolve("src", "raw_types_kysely");
180
- definitionFilePath = path.join(definitionsDir, `${dataset.database}.ts`);
181
- execSync(`kysely-codegen --url ${dbConnectionString} --schema=${dataset.schema} --out-file=${definitionFilePath}`, {
182
- // cwd: dataDir,
183
- env: process.env,
184
- encoding: "utf-8",
185
- // stdio: ["ignore", "ignore", "pipe"],
186
- });
187
- }
188
- }
189
- async function retrieveOpenData() {
190
- const dataDir = options.dataDir;
191
- assert(dataDir, "Missing argument: data directory");
192
- process.env = {
193
- ...process.env,
194
- PGHOST: process.env.PGHOST || config.db.host,
195
- PGPORT: process.env.PGPORT || config.db.port,
196
- PGUSER: process.env.PGUSER || config.db.user,
197
- PGPASSWORD: process.env.PGPASSWORD || config.db.password,
198
- };
199
- assert(process.env.PGHOST
200
- && process.env.PGPORT
201
- && process.env.PGUSER
202
- && process.env.PGPASSWORD, "Missing database configuration: environment variables PGHOST, PGPORT, PGUSER and PGPASSWORD or TRICOTEUSES_SENAT_DB_* in .env file");
203
- const choosenDatasets = getChosenFromEnabledDatasets(options.categories);
204
- // await Promise.all(choosenDatasets.map(dataset => retrieveDataset(dataDir, dataset)))
205
- for (const dataset of choosenDatasets) {
206
- await retrieveDataset(dataDir, dataset);
207
- }
208
- }
209
- retrieveOpenData()
210
- .then(() => process.exit(0))
211
- .catch((error) => {
212
- console.log(error);
213
- process.exit(1);
214
- });
@@ -1 +0,0 @@
1
- export {};