@tricoteuses/senat 3.1.0 → 3.1.2

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 (306) hide show
  1. package/lib/src/loaders.d.ts +3 -3
  2. package/lib/src/loaders.js +1 -1
  3. package/lib/src/model/agenda.d.ts +1 -1
  4. package/lib/src/model/commission.d.ts +2 -2
  5. package/lib/src/model/seance.d.ts +1 -1
  6. package/lib/src/types/ameli.d.ts +4 -1761
  7. package/lib/src/types/ameli.js +1 -1074
  8. package/lib/src/types/debats.d.ts +2 -380
  9. package/lib/src/types/debats.js +1 -266
  10. package/lib/src/types/dosleg.d.ts +69 -2953
  11. package/lib/src/types/dosleg.js +1 -2005
  12. package/lib/src/types/questions.d.ts +2 -699
  13. package/lib/src/types/questions.js +1 -493
  14. package/lib/src/types/sens.d.ts +7 -7842
  15. package/lib/src/types/sens.js +1 -4691
  16. package/lib/src/utils/nvs-parsing.d.ts +1 -1
  17. package/lib/src/utils/nvs-parsing.js +9 -1
  18. package/lib/src/videos/pipeline.d.ts +3 -3
  19. package/lib/src/videos/pipeline.js +2 -2
  20. package/package.json +1 -1
  21. package/lib/add-js-extensions-v2.d.ts +0 -1
  22. package/lib/add-js-extensions-v2.js +0 -23
  23. package/lib/add-js-extensions.d.ts +0 -1
  24. package/lib/add-js-extensions.js +0 -17
  25. package/lib/aggregates.d.ts +0 -52
  26. package/lib/aggregates.js +0 -930
  27. package/lib/aggregates.mjs +0 -713
  28. package/lib/aggregates.ts +0 -833
  29. package/lib/config.d.ts +0 -10
  30. package/lib/config.js +0 -16
  31. package/lib/config.mjs +0 -16
  32. package/lib/config.ts +0 -26
  33. package/lib/databases.d.ts +0 -2
  34. package/lib/databases.js +0 -26
  35. package/lib/databases.mjs +0 -57
  36. package/lib/databases.ts +0 -71
  37. package/lib/datasets.d.ts +0 -34
  38. package/lib/datasets.js +0 -233
  39. package/lib/datasets.mjs +0 -78
  40. package/lib/datasets.ts +0 -118
  41. package/lib/fields.d.ts +0 -10
  42. package/lib/fields.js +0 -68
  43. package/lib/fields.mjs +0 -22
  44. package/lib/fields.ts +0 -29
  45. package/lib/git.d.ts +0 -26
  46. package/lib/git.js +0 -167
  47. package/lib/index.d.ts +0 -13
  48. package/lib/index.js +0 -1
  49. package/lib/index.mjs +0 -7
  50. package/lib/index.ts +0 -64
  51. package/lib/inserters.d.ts +0 -98
  52. package/lib/inserters.js +0 -500
  53. package/lib/inserters.mjs +0 -360
  54. package/lib/inserters.ts +0 -521
  55. package/lib/legislatures.json +0 -38
  56. package/lib/loaders.d.ts +0 -58
  57. package/lib/loaders.js +0 -286
  58. package/lib/loaders.mjs +0 -158
  59. package/lib/loaders.ts +0 -271
  60. package/lib/model/agenda.d.ts +0 -6
  61. package/lib/model/agenda.js +0 -148
  62. package/lib/model/ameli.d.ts +0 -51
  63. package/lib/model/ameli.js +0 -149
  64. package/lib/model/ameli.mjs +0 -84
  65. package/lib/model/ameli.ts +0 -100
  66. package/lib/model/commission.d.ts +0 -18
  67. package/lib/model/commission.js +0 -269
  68. package/lib/model/debats.d.ts +0 -67
  69. package/lib/model/debats.js +0 -95
  70. package/lib/model/debats.mjs +0 -43
  71. package/lib/model/debats.ts +0 -68
  72. package/lib/model/documents.d.ts +0 -12
  73. package/lib/model/documents.js +0 -151
  74. package/lib/model/dosleg.d.ts +0 -7
  75. package/lib/model/dosleg.js +0 -326
  76. package/lib/model/dosleg.mjs +0 -196
  77. package/lib/model/dosleg.ts +0 -240
  78. package/lib/model/index.d.ts +0 -7
  79. package/lib/model/index.js +0 -7
  80. package/lib/model/index.mjs +0 -5
  81. package/lib/model/index.ts +0 -15
  82. package/lib/model/questions.d.ts +0 -45
  83. package/lib/model/questions.js +0 -89
  84. package/lib/model/questions.mjs +0 -71
  85. package/lib/model/questions.ts +0 -93
  86. package/lib/model/scrutins.d.ts +0 -13
  87. package/lib/model/scrutins.js +0 -114
  88. package/lib/model/seance.d.ts +0 -3
  89. package/lib/model/seance.js +0 -267
  90. package/lib/model/sens.d.ts +0 -146
  91. package/lib/model/sens.js +0 -454
  92. package/lib/model/sens.mjs +0 -415
  93. package/lib/model/sens.ts +0 -516
  94. package/lib/model/texte.d.ts +0 -7
  95. package/lib/model/texte.js +0 -256
  96. package/lib/model/texte.mjs +0 -208
  97. package/lib/model/texte.ts +0 -229
  98. package/lib/model/util.d.ts +0 -9
  99. package/lib/model/util.js +0 -38
  100. package/lib/model/util.mjs +0 -19
  101. package/lib/model/util.ts +0 -32
  102. package/lib/parsers/texte.d.ts +0 -7
  103. package/lib/parsers/texte.js +0 -228
  104. package/lib/raw_types/ameli.d.ts +0 -914
  105. package/lib/raw_types/ameli.js +0 -5
  106. package/lib/raw_types/ameli.mjs +0 -163
  107. package/lib/raw_types/debats.d.ts +0 -207
  108. package/lib/raw_types/debats.js +0 -5
  109. package/lib/raw_types/debats.mjs +0 -58
  110. package/lib/raw_types/dosleg.d.ts +0 -1619
  111. package/lib/raw_types/dosleg.js +0 -5
  112. package/lib/raw_types/dosleg.mjs +0 -438
  113. package/lib/raw_types/questions.d.ts +0 -419
  114. package/lib/raw_types/questions.js +0 -5
  115. package/lib/raw_types/questions.mjs +0 -11
  116. package/lib/raw_types/senat.d.ts +0 -11368
  117. package/lib/raw_types/senat.js +0 -5
  118. package/lib/raw_types/sens.d.ts +0 -8248
  119. package/lib/raw_types/sens.js +0 -5
  120. package/lib/raw_types/sens.mjs +0 -508
  121. package/lib/raw_types_kysely/ameli.d.ts +0 -915
  122. package/lib/raw_types_kysely/ameli.js +0 -7
  123. package/lib/raw_types_kysely/ameli.mjs +0 -5
  124. package/lib/raw_types_kysely/ameli.ts +0 -951
  125. package/lib/raw_types_kysely/debats.d.ts +0 -207
  126. package/lib/raw_types_kysely/debats.js +0 -7
  127. package/lib/raw_types_kysely/debats.mjs +0 -5
  128. package/lib/raw_types_kysely/debats.ts +0 -222
  129. package/lib/raw_types_kysely/dosleg.d.ts +0 -3532
  130. package/lib/raw_types_kysely/dosleg.js +0 -7
  131. package/lib/raw_types_kysely/dosleg.mjs +0 -5
  132. package/lib/raw_types_kysely/dosleg.ts +0 -3621
  133. package/lib/raw_types_kysely/questions.d.ts +0 -414
  134. package/lib/raw_types_kysely/questions.js +0 -7
  135. package/lib/raw_types_kysely/questions.mjs +0 -5
  136. package/lib/raw_types_kysely/questions.ts +0 -426
  137. package/lib/raw_types_kysely/sens.d.ts +0 -4394
  138. package/lib/raw_types_kysely/sens.js +0 -7
  139. package/lib/raw_types_kysely/sens.mjs +0 -5
  140. package/lib/raw_types_kysely/sens.ts +0 -4499
  141. package/lib/raw_types_schemats/ameli.d.ts +0 -539
  142. package/lib/raw_types_schemats/ameli.js +0 -2
  143. package/lib/raw_types_schemats/ameli.mjs +0 -2
  144. package/lib/raw_types_schemats/ameli.ts +0 -601
  145. package/lib/raw_types_schemats/debats.d.ts +0 -127
  146. package/lib/raw_types_schemats/debats.js +0 -2
  147. package/lib/raw_types_schemats/debats.mjs +0 -2
  148. package/lib/raw_types_schemats/debats.ts +0 -145
  149. package/lib/raw_types_schemats/dosleg.d.ts +0 -977
  150. package/lib/raw_types_schemats/dosleg.js +0 -2
  151. package/lib/raw_types_schemats/dosleg.mjs +0 -2
  152. package/lib/raw_types_schemats/dosleg.ts +0 -2193
  153. package/lib/raw_types_schemats/questions.d.ts +0 -235
  154. package/lib/raw_types_schemats/questions.js +0 -2
  155. package/lib/raw_types_schemats/questions.mjs +0 -2
  156. package/lib/raw_types_schemats/questions.ts +0 -249
  157. package/lib/raw_types_schemats/sens.d.ts +0 -6915
  158. package/lib/raw_types_schemats/sens.js +0 -2
  159. package/lib/raw_types_schemats/sens.mjs +0 -2
  160. package/lib/raw_types_schemats/sens.ts +0 -2907
  161. package/lib/scripts/convert_data.d.ts +0 -1
  162. package/lib/scripts/convert_data.js +0 -354
  163. package/lib/scripts/convert_data.mjs +0 -181
  164. package/lib/scripts/convert_data.ts +0 -243
  165. package/lib/scripts/data-download.d.ts +0 -1
  166. package/lib/scripts/data-download.js +0 -12
  167. package/lib/scripts/datautil.d.ts +0 -8
  168. package/lib/scripts/datautil.js +0 -34
  169. package/lib/scripts/datautil.mjs +0 -16
  170. package/lib/scripts/datautil.ts +0 -19
  171. package/lib/scripts/images/transparent_150x192.jpg +0 -0
  172. package/lib/scripts/images/transparent_155x225.jpg +0 -0
  173. package/lib/scripts/parse_textes.d.ts +0 -1
  174. package/lib/scripts/parse_textes.js +0 -44
  175. package/lib/scripts/parse_textes.mjs +0 -46
  176. package/lib/scripts/parse_textes.ts +0 -65
  177. package/lib/scripts/retrieve_agenda.d.ts +0 -1
  178. package/lib/scripts/retrieve_agenda.js +0 -132
  179. package/lib/scripts/retrieve_cr_commission.d.ts +0 -1
  180. package/lib/scripts/retrieve_cr_commission.js +0 -364
  181. package/lib/scripts/retrieve_cr_seance.d.ts +0 -6
  182. package/lib/scripts/retrieve_cr_seance.js +0 -347
  183. package/lib/scripts/retrieve_documents.d.ts +0 -3
  184. package/lib/scripts/retrieve_documents.js +0 -219
  185. package/lib/scripts/retrieve_documents.mjs +0 -249
  186. package/lib/scripts/retrieve_documents.ts +0 -298
  187. package/lib/scripts/retrieve_open_data.d.ts +0 -1
  188. package/lib/scripts/retrieve_open_data.js +0 -315
  189. package/lib/scripts/retrieve_open_data.mjs +0 -217
  190. package/lib/scripts/retrieve_open_data.ts +0 -268
  191. package/lib/scripts/retrieve_senateurs_photos.d.ts +0 -1
  192. package/lib/scripts/retrieve_senateurs_photos.js +0 -147
  193. package/lib/scripts/retrieve_senateurs_photos.mjs +0 -147
  194. package/lib/scripts/retrieve_senateurs_photos.ts +0 -177
  195. package/lib/scripts/retrieve_videos.d.ts +0 -1
  196. package/lib/scripts/retrieve_videos.js +0 -461
  197. package/lib/scripts/shared/cli_helpers.d.ts +0 -95
  198. package/lib/scripts/shared/cli_helpers.js +0 -91
  199. package/lib/scripts/shared/cli_helpers.ts +0 -36
  200. package/lib/scripts/shared/util.d.ts +0 -4
  201. package/lib/scripts/shared/util.js +0 -35
  202. package/lib/scripts/shared/util.ts +0 -33
  203. package/lib/scripts/test_iter_load.d.ts +0 -1
  204. package/lib/scripts/test_iter_load.js +0 -12
  205. package/lib/src/ameli.d.ts +0 -66
  206. package/lib/src/ameli.js +0 -1
  207. package/lib/src/databases.d.ts +0 -3
  208. package/lib/src/databases.js +0 -26
  209. package/lib/src/db_types/ameli.d.ts +0 -1762
  210. package/lib/src/db_types/ameli.js +0 -1074
  211. package/lib/src/db_types/debats.d.ts +0 -380
  212. package/lib/src/db_types/debats.js +0 -266
  213. package/lib/src/db_types/dosleg.d.ts +0 -2954
  214. package/lib/src/db_types/dosleg.js +0 -2005
  215. package/lib/src/db_types/questions.d.ts +0 -699
  216. package/lib/src/db_types/questions.js +0 -493
  217. package/lib/src/db_types/sens.d.ts +0 -7843
  218. package/lib/src/db_types/sens.js +0 -4691
  219. package/lib/src/debats.d.ts +0 -38
  220. package/lib/src/debats.js +0 -1
  221. package/lib/src/dosleg.d.ts +0 -142
  222. package/lib/src/dosleg.js +0 -193
  223. package/lib/src/model/ameli_postgres.d.ts +0 -67
  224. package/lib/src/model/ameli_postgres.js +0 -150
  225. package/lib/src/other_types/questions.d.ts +0 -2
  226. package/lib/src/other_types/questions.js +0 -1
  227. package/lib/src/questions.d.ts +0 -53
  228. package/lib/src/questions.js +0 -1
  229. package/lib/src/raw_types/senat.d.ts +0 -11372
  230. package/lib/src/raw_types/senat.js +0 -5
  231. package/lib/src/rich_types/agenda.d.ts +0 -45
  232. package/lib/src/rich_types/agenda.js +0 -1
  233. package/lib/src/rich_types/compte_rendu.d.ts +0 -83
  234. package/lib/src/rich_types/compte_rendu.js +0 -1
  235. package/lib/src/rich_types/sessions.d.ts +0 -6
  236. package/lib/src/rich_types/sessions.js +0 -19
  237. package/lib/src/rich_types/texte.d.ts +0 -72
  238. package/lib/src/rich_types/texte.js +0 -15
  239. package/lib/src/scripts/test_iter_load.d.ts +0 -1
  240. package/lib/src/scripts/test_iter_load.js +0 -12
  241. package/lib/src/sens.d.ts +0 -104
  242. package/lib/src/sens.js +0 -1
  243. package/lib/strings.d.ts +0 -1
  244. package/lib/strings.js +0 -18
  245. package/lib/strings.mjs +0 -18
  246. package/lib/strings.ts +0 -26
  247. package/lib/tsconfig.tsbuildinfo +0 -1
  248. package/lib/types/agenda.d.ts +0 -44
  249. package/lib/types/agenda.js +0 -1
  250. package/lib/types/ameli.d.ts +0 -5
  251. package/lib/types/ameli.js +0 -1
  252. package/lib/types/ameli.mjs +0 -13
  253. package/lib/types/ameli.ts +0 -21
  254. package/lib/types/compte_rendu.d.ts +0 -83
  255. package/lib/types/compte_rendu.js +0 -1
  256. package/lib/types/debats.d.ts +0 -2
  257. package/lib/types/debats.js +0 -1
  258. package/lib/types/debats.mjs +0 -2
  259. package/lib/types/debats.ts +0 -6
  260. package/lib/types/dosleg.d.ts +0 -70
  261. package/lib/types/dosleg.js +0 -1
  262. package/lib/types/dosleg.mjs +0 -151
  263. package/lib/types/dosleg.ts +0 -284
  264. package/lib/types/questions.d.ts +0 -2
  265. package/lib/types/questions.js +0 -1
  266. package/lib/types/questions.mjs +0 -1
  267. package/lib/types/questions.ts +0 -3
  268. package/lib/types/sens.d.ts +0 -10
  269. package/lib/types/sens.js +0 -1
  270. package/lib/types/sens.mjs +0 -1
  271. package/lib/types/sens.ts +0 -12
  272. package/lib/types/sessions.d.ts +0 -5
  273. package/lib/types/sessions.js +0 -84
  274. package/lib/types/sessions.mjs +0 -43
  275. package/lib/types/sessions.ts +0 -42
  276. package/lib/types/texte.d.ts +0 -74
  277. package/lib/types/texte.js +0 -16
  278. package/lib/types/texte.mjs +0 -16
  279. package/lib/types/texte.ts +0 -76
  280. package/lib/typings/windows-1252.d.js +0 -2
  281. package/lib/typings/windows-1252.d.mjs +0 -2
  282. package/lib/typings/windows-1252.d.ts +0 -11
  283. package/lib/utils/cr_spliting.d.ts +0 -28
  284. package/lib/utils/cr_spliting.js +0 -265
  285. package/lib/utils/date.d.ts +0 -10
  286. package/lib/utils/date.js +0 -100
  287. package/lib/utils/nvs-timecode.d.ts +0 -7
  288. package/lib/utils/nvs-timecode.js +0 -79
  289. package/lib/utils/reunion_grouping.d.ts +0 -9
  290. package/lib/utils/reunion_grouping.js +0 -361
  291. package/lib/utils/reunion_odj_building.d.ts +0 -5
  292. package/lib/utils/reunion_odj_building.js +0 -154
  293. package/lib/utils/reunion_parsing.d.ts +0 -23
  294. package/lib/utils/reunion_parsing.js +0 -209
  295. package/lib/utils/scoring.d.ts +0 -14
  296. package/lib/utils/scoring.js +0 -147
  297. package/lib/utils/string_cleaning.d.ts +0 -7
  298. package/lib/utils/string_cleaning.js +0 -57
  299. package/lib/validators/config.d.ts +0 -9
  300. package/lib/validators/config.js +0 -10
  301. package/lib/validators/config.mjs +0 -54
  302. package/lib/validators/config.ts +0 -79
  303. package/lib/validators/senat.d.ts +0 -0
  304. package/lib/validators/senat.js +0 -28
  305. package/lib/validators/senat.mjs +0 -24
  306. package/lib/validators/senat.ts +0 -26
@@ -1,256 +0,0 @@
1
- import { JSDOM } from "jsdom";
2
- import { AKN_IDENTIFICATION_STRUCTURE_REGEXP, AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP } from "../scripts/datautil";
3
- import { DivisionType, } from "../types/texte";
4
- function buildWorklow(metaElement) {
5
- const stepElements = metaElement.querySelectorAll("workflow step");
6
- const steps = [];
7
- for (const stepElement of stepElements) {
8
- const identification = stepElement.getAttribute("href") ?? "";
9
- const identificationParts = AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups;
10
- steps.push({
11
- eId: stepElement.getAttribute("eId"),
12
- date: stepElement.getAttribute("date")
13
- ? new Date(stepElement.getAttribute("date") ?? "")
14
- : null,
15
- type: identificationParts?.["type"] || null,
16
- session: identificationParts?.["session"] || null,
17
- numero: identificationParts?.["numTexte"] || null,
18
- version: identificationParts?.["version"]
19
- ? identificationParts["version"]
20
- : null,
21
- outcome: stepElement.getAttribute("outcome"),
22
- });
23
- }
24
- return steps;
25
- }
26
- function buildDivision(node, index) {
27
- const eId = node.getAttribute("eId");
28
- const tag = node.nodeName;
29
- const level = DivisionType[tag];
30
- const titleNode = node.querySelector("num");
31
- const subtitleNode = node.querySelector("heading");
32
- const headings = [
33
- ...(titleNode
34
- ? [
35
- {
36
- text: titleNode.textContent?.trim() ?? null,
37
- html: titleNode.innerHTML?.trim() ?? null,
38
- },
39
- ]
40
- : []),
41
- ...(subtitleNode
42
- ? [
43
- {
44
- text: subtitleNode.textContent?.trim() ?? null,
45
- html: subtitleNode.innerHTML?.trim() ?? null,
46
- },
47
- ]
48
- : []),
49
- ];
50
- const division = {
51
- index,
52
- eId,
53
- tag,
54
- level,
55
- headings,
56
- };
57
- if (tag === "article") {
58
- ;
59
- division.alineas = [];
60
- }
61
- return division;
62
- }
63
- function buildAlinea(contentNode, alineaNode) {
64
- const eId = alineaNode.getAttribute("eId");
65
- const heading = {
66
- text: alineaNode.querySelector("num")?.textContent ?? null,
67
- };
68
- const pastille = alineaNode.getAttribute("data:pastille") ?? null;
69
- return {
70
- eId,
71
- heading,
72
- text: contentNode.textContent?.trim() ?? null,
73
- html: contentNode.innerHTML?.trim() ?? null,
74
- pastille,
75
- };
76
- }
77
- function buildEmptyArticle(index) {
78
- return {
79
- index: index,
80
- eId: "",
81
- tag: "article",
82
- level: DivisionType["article"],
83
- headings: [],
84
- alineas: [],
85
- };
86
- }
87
- function flattenTexte(texteContentRoot) {
88
- const divisions = [];
89
- let divisionIndex = 0;
90
- const iter = (node) => {
91
- if (node.nodeName === "content") {
92
- return;
93
- }
94
- switch (node.nodeName) {
95
- case "tome":
96
- case "part":
97
- case "book":
98
- case "title":
99
- case "subtitle":
100
- case "chapter":
101
- case "section":
102
- case "subsection":
103
- case "paragraph":
104
- case "article":
105
- divisions.push(buildDivision(node, divisionIndex++));
106
- break;
107
- }
108
- if (node.nodeName === "alinea") {
109
- Array.from(node.childNodes)
110
- // Find direct content children programmatically
111
- // because `:scope` selector does not work
112
- // https://github.com/jsdom/jsdom/issues/2998
113
- .filter((alineaChildNode) => alineaChildNode.nodeName === "content")
114
- .forEach((alineaContentNode) => {
115
- // Hypothesis: alineas should always be enclosed in articles
116
- let lastArticle = divisions.findLast((division) => division.tag === "article");
117
- if (!lastArticle) {
118
- lastArticle = buildEmptyArticle(divisionIndex++);
119
- divisions.push(lastArticle);
120
- }
121
- lastArticle.alineas.push(buildAlinea(alineaContentNode, node));
122
- });
123
- }
124
- if (node.hasChildNodes()) {
125
- node.childNodes.forEach((childNode) => iter(childNode));
126
- }
127
- };
128
- iter(texteContentRoot);
129
- return divisions;
130
- }
131
- export function transformTexte(document) {
132
- const metaElement = document.querySelector("meta");
133
- const preambleElement = document.querySelector("preamble");
134
- const identification = metaElement
135
- ?.querySelector("FRBRExpression FRBRuri")
136
- ?.getAttribute("value") ?? "";
137
- const identificationParts = AKN_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups;
138
- const bodyElement = document.querySelector("body");
139
- const sessionYears = identificationParts?.["session"]?.split("-") || null;
140
- const datePresentation = metaElement
141
- ?.querySelector("FRBRdate[name='#presentation']")
142
- ?.getAttribute("date");
143
- const dateDepot = metaElement
144
- ?.querySelector("FRBRdate[name='#depot']")
145
- ?.getAttribute("date");
146
- const datePublicationXml = metaElement
147
- ?.querySelector("FRBRdate[name='#publication-xml']")
148
- ?.getAttribute("date");
149
- return {
150
- titre: preambleElement?.querySelector("docTitle")?.textContent || null,
151
- titreCourt: metaElement
152
- ?.querySelector("FRBRalias[name='intitule-court']")
153
- ?.getAttribute("value") || null,
154
- signetDossier: metaElement
155
- ?.querySelector("FRBRalias[name='signet-dossier-legislatif-senat']")
156
- ?.getAttribute("value") || null,
157
- urlDossierSenat: metaElement
158
- ?.querySelector("FRBRalias[name='url-senat']")
159
- ?.getAttribute("value") || null,
160
- urlDossierAssemblee: metaElement
161
- ?.querySelector("FRBRalias[name='url-AN']")
162
- ?.getAttribute("value") || null,
163
- type: identificationParts?.["type"] || null,
164
- session: sessionYears && sessionYears.length > 0 ? sessionYears[0] : null,
165
- numero: identificationParts?.["numTexte"]
166
- ? parseInt(identificationParts["numTexte"])
167
- : null,
168
- datePresentation: datePresentation ? new Date(datePresentation) : null,
169
- dateDepot: dateDepot ? new Date(dateDepot) : null,
170
- datePublicationXml: datePublicationXml
171
- ? new Date(datePublicationXml)
172
- : null,
173
- version: identificationParts?.["version"]
174
- ? identificationParts["version"]
175
- : null,
176
- workflow: metaElement ? buildWorklow(metaElement) : [],
177
- divisions: bodyElement ? flattenTexte(bodyElement) : [],
178
- };
179
- }
180
- export function transformExposeDesMotifs(document) {
181
- const sectionElements = document.querySelectorAll("section");
182
- const exposeDesMotifsRegexp = new RegExp("EXPOS.{1,2}[\\n\\s]DES[\\n\\s]MOTIFS");
183
- for (const sectionElement of sectionElements) {
184
- const firstParagraph = sectionElement.querySelector("p:first-of-type");
185
- const secondParagraph = sectionElement.querySelector("p:nth-of-type(2)");
186
- if (!firstParagraph) {
187
- continue;
188
- }
189
- const firstParagraphContent = firstParagraph.textContent;
190
- const secondParagraphContent = secondParagraph?.textContent;
191
- if (!firstParagraphContent ||
192
- !exposeDesMotifsRegexp.test(firstParagraphContent.toUpperCase())) {
193
- if (!secondParagraphContent ||
194
- !exposeDesMotifsRegexp.test(secondParagraphContent.toUpperCase())) {
195
- continue;
196
- }
197
- else {
198
- secondParagraph.remove();
199
- }
200
- }
201
- firstParagraph.remove();
202
- return {
203
- text: sectionElement.textContent?.trim() ?? null,
204
- html: sectionElement.innerHTML?.trim() ?? null,
205
- };
206
- }
207
- return null;
208
- }
209
- export function parseTexte(texteXml) {
210
- try {
211
- const { document } = new JSDOM(texteXml, {
212
- contentType: "text/xml",
213
- }).window;
214
- return transformTexte(document);
215
- }
216
- catch (error) {
217
- console.error(`Could not parse texte with error ${error}`);
218
- }
219
- return null;
220
- }
221
- // Prevent from memory leak
222
- // https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
223
- export async function parseTexteFromFile(xmlFilePath) {
224
- try {
225
- const { document } = (await JSDOM.fromFile(xmlFilePath, { contentType: "text/xml" })).window;
226
- return transformTexte(document);
227
- }
228
- catch (error) {
229
- console.error(`Could not parse texte with error ${error}`);
230
- }
231
- return null;
232
- }
233
- export function parseExposeDesMotifs(exposeDesMotifsHtml) {
234
- try {
235
- const { document } = new JSDOM(exposeDesMotifsHtml, {
236
- contentType: "text/html",
237
- }).window;
238
- return transformExposeDesMotifs(document);
239
- }
240
- catch (error) {
241
- console.error(`Could not parse exposé des motifs with error ${error}`);
242
- }
243
- return null;
244
- }
245
- // Prevent from memory leak
246
- // https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
247
- export async function parseExposeDesMotifsFromFile(htmlFilePath) {
248
- try {
249
- const { document } = (await JSDOM.fromFile(htmlFilePath, { contentType: "text/html" })).window;
250
- return transformExposeDesMotifs(document);
251
- }
252
- catch (error) {
253
- console.error(`Could not parse exposé des motifs with error ${error}`);
254
- }
255
- return null;
256
- }
@@ -1,208 +0,0 @@
1
- import { JSDOM } from "jsdom";
2
- import { AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP, AKN_IDENTIFICATION_STRUCTURE_REGEXP } from "../scripts/datautil";
3
- import { DivisionType, } from "../types/texte";
4
- function buildWorklow(metaElement) {
5
- const stepElements = metaElement.querySelectorAll("workflow step");
6
- const steps = [];
7
- for (const stepElement of stepElements) {
8
- const identification = stepElement.getAttribute("href") ?? "";
9
- const identificationParts = AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups;
10
- steps.push({
11
- eId: stepElement.getAttribute("eId"),
12
- date: stepElement.getAttribute("date") ? new Date(stepElement.getAttribute("date") ?? "") : null,
13
- type: identificationParts?.type || null,
14
- session: identificationParts?.session || null,
15
- numero: identificationParts?.numTexte || null,
16
- version: identificationParts?.version ? identificationParts.version : null,
17
- outcome: stepElement.getAttribute("outcome"),
18
- });
19
- }
20
- return steps;
21
- }
22
- function buildDivision(node, index) {
23
- const eId = node.getAttribute("eId");
24
- const tag = node.nodeName;
25
- const level = DivisionType[tag];
26
- const titleNode = node.querySelector("num");
27
- const subtitleNode = node.querySelector("heading");
28
- const headings = [
29
- ...(titleNode ? [{
30
- text: titleNode.textContent?.trim() ?? null,
31
- html: titleNode.innerHTML?.trim() ?? null,
32
- }] : []),
33
- ...(subtitleNode ? [{
34
- text: subtitleNode.textContent?.trim() ?? null,
35
- html: subtitleNode.innerHTML?.trim() ?? null,
36
- }] : []),
37
- ];
38
- const division = {
39
- index,
40
- eId,
41
- tag,
42
- level,
43
- headings,
44
- };
45
- if (tag === "article") {
46
- division.alineas = [];
47
- }
48
- return division;
49
- }
50
- function buildAlinea(contentNode, alineaNode) {
51
- const eId = alineaNode.getAttribute("eId");
52
- const heading = {
53
- text: alineaNode.querySelector("num")?.textContent ?? null,
54
- };
55
- const pastille = alineaNode.getAttribute("data:pastille") ?? null;
56
- return {
57
- eId,
58
- heading,
59
- text: contentNode.textContent?.trim() ?? null,
60
- html: contentNode.innerHTML?.trim() ?? null,
61
- pastille,
62
- };
63
- }
64
- function buildEmptyArticle(index) {
65
- return {
66
- index: index,
67
- eId: "",
68
- tag: "article",
69
- level: DivisionType["article"],
70
- headings: [],
71
- alineas: [],
72
- };
73
- }
74
- function flattenTexte(texteContentRoot) {
75
- const divisions = [];
76
- let divisionIndex = 0;
77
- const iter = (node) => {
78
- if (node.nodeName === "content") {
79
- return;
80
- }
81
- switch (node.nodeName) {
82
- case "tome":
83
- case "part":
84
- case "book":
85
- case "title":
86
- case "subtitle":
87
- case "chapter":
88
- case "section":
89
- case "subsection":
90
- case "paragraph":
91
- case "article":
92
- divisions.push(buildDivision(node, divisionIndex++));
93
- break;
94
- }
95
- if (node.nodeName === "alinea") {
96
- Array.from(node.childNodes)
97
- // Find direct content children programmatically
98
- // because `:scope` selector does not work
99
- // https://github.com/jsdom/jsdom/issues/2998
100
- .filter((alineaChildNode) => alineaChildNode.nodeName === "content")
101
- .forEach((alineaContentNode) => {
102
- // Hypothesis: alineas should always be enclosed in articles
103
- let lastArticle = divisions.findLast(division => division.tag === "article");
104
- if (!lastArticle) {
105
- lastArticle = buildEmptyArticle(divisionIndex++);
106
- divisions.push(lastArticle);
107
- }
108
- lastArticle.alineas.push(buildAlinea(alineaContentNode, node));
109
- });
110
- }
111
- if (node.hasChildNodes()) {
112
- node.childNodes.forEach((childNode) => iter(childNode));
113
- }
114
- };
115
- iter(texteContentRoot);
116
- return divisions;
117
- }
118
- export function transformTexte(document) {
119
- const metaElement = document.querySelector("meta");
120
- const preambleElement = document.querySelector("preamble");
121
- const identification = metaElement?.querySelector("FRBRExpression FRBRuri")?.getAttribute("value") ?? "";
122
- const identificationParts = AKN_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups;
123
- const bodyElement = document.querySelector("body");
124
- const datePresentation = metaElement?.querySelector("FRBRdate[name='#presentation']")?.getAttribute("date");
125
- const dateDepot = metaElement?.querySelector("FRBRdate[name='#depot']")?.getAttribute("date");
126
- const datePublicationXml = metaElement?.querySelector("FRBRdate[name='#publication-xml']")?.getAttribute("date");
127
- return {
128
- titre: preambleElement?.querySelector("docTitle")?.textContent || null,
129
- titreCourt: metaElement?.querySelector("FRBRalias[name='intitule-court']")
130
- ?.getAttribute("value") || null,
131
- signetDossier: metaElement?.querySelector("FRBRalias[name='signet-dossier-legislatif-senat']")
132
- ?.getAttribute("value") || null,
133
- urlDossierSenat: metaElement?.querySelector("FRBRalias[name='url-senat']")?.getAttribute("value") || null,
134
- urlDossierAssemblee: metaElement?.querySelector("FRBRalias[name='url-AN']")?.getAttribute("value") || null,
135
- type: identificationParts?.type || null,
136
- session: identificationParts?.session || null,
137
- numero: identificationParts?.numTexte ? parseInt(identificationParts.numTexte) : null,
138
- datePresentation: datePresentation ? new Date(datePresentation) : null,
139
- dateDepot: dateDepot ? new Date(dateDepot) : null,
140
- datePublicationXml: datePublicationXml ? new Date(datePublicationXml) : null,
141
- version: identificationParts?.version ? identificationParts.version : null,
142
- workflow: metaElement ? buildWorklow(metaElement) : [],
143
- divisions: bodyElement ? flattenTexte(bodyElement) : [],
144
- };
145
- }
146
- export function transformExposeDesMotifs(document) {
147
- const sectionElements = document.querySelectorAll("section");
148
- const exposeDesMotifsRegexp = new RegExp("EXPOS.{1,2}[\\n\\s]DES[\\n\\s]MOTIFS");
149
- for (const sectionElement of sectionElements) {
150
- const firstParagraph = sectionElement.querySelector("p:first-of-type");
151
- const secondParagraph = sectionElement.querySelector("p:nth-of-type(2)");
152
- if (!firstParagraph) {
153
- continue;
154
- }
155
- const firstParagraphContent = firstParagraph.textContent;
156
- const secondParagraphContent = secondParagraph?.textContent;
157
- if (!firstParagraphContent || !exposeDesMotifsRegexp.test(firstParagraphContent.toUpperCase())) {
158
- if (!secondParagraphContent || !exposeDesMotifsRegexp.test(secondParagraphContent.toUpperCase())) {
159
- continue;
160
- }
161
- else {
162
- sectionElement.removeChild(secondParagraph);
163
- }
164
- }
165
- sectionElement.removeChild(firstParagraph);
166
- return {
167
- text: sectionElement.textContent?.trim() ?? null,
168
- html: sectionElement.innerHTML?.trim() ?? null,
169
- };
170
- }
171
- return null;
172
- }
173
- export function parseTexte(texteXml) {
174
- try {
175
- const { document } = (new JSDOM(texteXml, {
176
- contentType: "text/xml",
177
- })).window;
178
- return transformTexte(document);
179
- }
180
- catch (error) {
181
- console.error(`Could not parse texte with error ${error}`);
182
- }
183
- return null;
184
- }
185
- // Prevent from memory leak
186
- // https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
187
- export async function parseTexteFromFile(xmlFilePath) {
188
- try {
189
- const { document } = (await JSDOM.fromFile(xmlFilePath, { contentType: "text/xml" })).window;
190
- return transformTexte(document);
191
- }
192
- catch (error) {
193
- console.error(`Could not parse texte with error ${error}`);
194
- }
195
- return null;
196
- }
197
- export function parseExposeDesMotifs(exposeDesMotifsHtml) {
198
- const { document } = (new JSDOM(exposeDesMotifsHtml, {
199
- contentType: "text/html",
200
- })).window;
201
- return transformExposeDesMotifs(document);
202
- }
203
- // Prevent from memory leak
204
- // https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
205
- export async function parseExposeDesMotifsFromFile(htmlFilePath) {
206
- const { document } = (await JSDOM.fromFile(htmlFilePath, { contentType: "text/html" })).window;
207
- return transformExposeDesMotifs(document);
208
- }
@@ -1,229 +0,0 @@
1
- import { JSDOM } from "jsdom"
2
-
3
- import { AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP, AKN_IDENTIFICATION_STRUCTURE_REGEXP } from "../scripts/datautil"
4
- import {
5
- Alinea,
6
- Article,
7
- Division,
8
- DivisionContent,
9
- DivisionTag,
10
- DivisionType, ExposeDesMotifs,
11
- FlatTexte, Step,
12
- Version,
13
- } from "../types/texte"
14
-
15
- function buildWorklow (metaElement: HTMLMetaElement): Step[] {
16
- const stepElements = metaElement.querySelectorAll("workflow step")
17
- const steps: Step[] = []
18
- for (const stepElement of stepElements) {
19
- const identification = stepElement.getAttribute("href") ?? ""
20
- const identificationParts = AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups
21
- steps.push({
22
- eId: stepElement.getAttribute("eId")!,
23
- date: stepElement.getAttribute("date") ? new Date(stepElement.getAttribute("date") ?? "") : null,
24
- type: identificationParts?.type || null,
25
- session: identificationParts?.session || null,
26
- numero: identificationParts?.numTexte || null,
27
- version: identificationParts?.version ? identificationParts.version as Version : null,
28
- outcome: stepElement.getAttribute("outcome"),
29
- })
30
- }
31
- return steps
32
- }
33
-
34
- function buildDivision (node: Node, index: number): Division {
35
- const eId = (node as Element).getAttribute("eId")!
36
- const tag = node.nodeName as DivisionTag
37
- const level = DivisionType[tag]
38
- const titleNode = (node as Element).querySelector("num")
39
- const subtitleNode = (node as Element).querySelector("heading")
40
- const headings: DivisionContent[] = [
41
- ...(titleNode ? [{
42
- text: titleNode.textContent?.trim() ?? null,
43
- html: titleNode.innerHTML?.trim() ?? null,
44
- }] : []),
45
- ...(subtitleNode ? [{
46
- text: subtitleNode.textContent?.trim() ?? null,
47
- html: subtitleNode.innerHTML?.trim() ?? null,
48
- }] : []),
49
- ]
50
- const division = {
51
- index,
52
- eId,
53
- tag,
54
- level,
55
- headings,
56
- }
57
- if (tag === "article") {
58
- (division as Article).alineas = []
59
- }
60
- return division
61
- }
62
-
63
- function buildAlinea (contentNode: Node, alineaNode: Node): Alinea {
64
- const eId = (alineaNode as Element).getAttribute("eId")!
65
- const heading = {
66
- text: (alineaNode as Element).querySelector("num")?.textContent ?? null,
67
- }
68
- const pastille = (alineaNode as Element).getAttribute("data:pastille") ?? null
69
- return {
70
- eId,
71
- heading,
72
- text: (contentNode as Element).textContent?.trim() ?? null,
73
- html: (contentNode as Element).innerHTML?.trim() ?? null,
74
- pastille,
75
- }
76
- }
77
-
78
- function buildEmptyArticle (index: number): Article {
79
- return {
80
- index: index,
81
- eId: "",
82
- tag: "article",
83
- level: DivisionType["article"],
84
- headings: [],
85
- alineas: [],
86
- }
87
- }
88
-
89
- function flattenTexte (texteContentRoot: Node): Division[] {
90
- const divisions: Division[] = []
91
- let divisionIndex = 0
92
- const iter = (node: Node) => {
93
- if (node.nodeName === "content") {
94
- return
95
- }
96
- switch (node.nodeName) {
97
- case "tome":
98
- case "part":
99
- case "book":
100
- case "title":
101
- case "subtitle":
102
- case "chapter":
103
- case "section":
104
- case "subsection":
105
- case "paragraph":
106
- case "article":
107
- divisions.push(buildDivision(node, divisionIndex++))
108
- break
109
- }
110
- if (node.nodeName === "alinea") {
111
- Array.from(node.childNodes)
112
- // Find direct content children programmatically
113
- // because `:scope` selector does not work
114
- // https://github.com/jsdom/jsdom/issues/2998
115
- .filter((alineaChildNode: Node) => alineaChildNode.nodeName === "content")
116
- .forEach((alineaContentNode: Node) => {
117
- // Hypothesis: alineas should always be enclosed in articles
118
- let lastArticle = divisions.findLast(division => division.tag === "article") as Article
119
- if (!lastArticle) {
120
- lastArticle = buildEmptyArticle(divisionIndex++)
121
- divisions.push(lastArticle)
122
- }
123
- lastArticle.alineas.push(buildAlinea(alineaContentNode, node))
124
- })
125
- }
126
- if (node.hasChildNodes()) {
127
- node.childNodes.forEach((childNode: Node) => iter(childNode))
128
- }
129
- }
130
- iter(texteContentRoot)
131
- return divisions
132
- }
133
-
134
- export function transformTexte (document: Document): FlatTexte | null {
135
- const metaElement = document.querySelector("meta")
136
- const preambleElement = document.querySelector("preamble")
137
- const identification = metaElement?.querySelector("FRBRExpression FRBRuri")?.getAttribute("value") ?? ""
138
- const identificationParts = AKN_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups
139
- const bodyElement = document.querySelector("body")
140
- const datePresentation = metaElement?.querySelector("FRBRdate[name='#presentation']")?.getAttribute("date")
141
- const dateDepot = metaElement?.querySelector("FRBRdate[name='#depot']")?.getAttribute("date")
142
- const datePublicationXml = metaElement?.querySelector("FRBRdate[name='#publication-xml']")?.getAttribute("date")
143
- return {
144
- titre: preambleElement?.querySelector("docTitle")?.textContent || null,
145
- titreCourt: metaElement?.querySelector("FRBRalias[name='intitule-court']")
146
- ?.getAttribute("value") || null,
147
- signetDossier:
148
- metaElement?.querySelector("FRBRalias[name='signet-dossier-legislatif-senat']")
149
- ?.getAttribute("value") || null,
150
- urlDossierSenat: metaElement?.querySelector("FRBRalias[name='url-senat']")?.getAttribute("value") || null,
151
- urlDossierAssemblee: metaElement?.querySelector("FRBRalias[name='url-AN']")?.getAttribute("value") || null,
152
- type: identificationParts?.type || null,
153
- session: identificationParts?.session || null,
154
- numero: identificationParts?.numTexte ? parseInt(identificationParts.numTexte) : null,
155
- datePresentation: datePresentation ? new Date(datePresentation) : null,
156
- dateDepot: dateDepot ? new Date(dateDepot) : null,
157
- datePublicationXml: datePublicationXml ? new Date(datePublicationXml) : null,
158
- version: identificationParts?.version ? identificationParts.version as Version : null,
159
- workflow: metaElement ? buildWorklow(metaElement) : [],
160
- divisions: bodyElement ? flattenTexte(bodyElement) : [],
161
- }
162
- }
163
-
164
- export function transformExposeDesMotifs (document: Document): ExposeDesMotifs | null {
165
- const sectionElements = document.querySelectorAll("section")
166
- const exposeDesMotifsRegexp = new RegExp("EXPOS.{1,2}[\\n\\s]DES[\\n\\s]MOTIFS")
167
- for (const sectionElement of sectionElements) {
168
- const firstParagraph = sectionElement.querySelector("p:first-of-type")
169
- const secondParagraph = sectionElement.querySelector("p:nth-of-type(2)")
170
- if (!firstParagraph) {
171
- continue
172
- }
173
-
174
- const firstParagraphContent = firstParagraph.textContent
175
- const secondParagraphContent = secondParagraph?.textContent
176
- if (!firstParagraphContent || !exposeDesMotifsRegexp.test(firstParagraphContent.toUpperCase())) {
177
- if (!secondParagraphContent || !exposeDesMotifsRegexp.test(secondParagraphContent.toUpperCase())) {
178
- continue
179
- } else {
180
- sectionElement.removeChild(secondParagraph)
181
- }
182
- }
183
-
184
- sectionElement.removeChild(firstParagraph)
185
- return {
186
- text: sectionElement.textContent?.trim() ?? null,
187
- html: sectionElement.innerHTML?.trim() ?? null,
188
- }
189
- }
190
- return null
191
- }
192
-
193
- export function parseTexte (texteXml: string): FlatTexte | null {
194
- try {
195
- const { document } = (new JSDOM(texteXml, {
196
- contentType: "text/xml",
197
- })).window
198
- return transformTexte(document)
199
- } catch (error: any) {
200
- console.error(`Could not parse texte with error ${error}`)
201
- }
202
- return null
203
- }
204
-
205
- // Prevent from memory leak
206
- // https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
207
- export async function parseTexteFromFile (xmlFilePath: string): Promise<FlatTexte | null> {
208
- try {
209
- const { document } = (await JSDOM.fromFile(xmlFilePath, { contentType: "text/xml" })).window
210
- return transformTexte(document)
211
- } catch (error: any) {
212
- console.error(`Could not parse texte with error ${error}`)
213
- }
214
- return null
215
- }
216
-
217
- export function parseExposeDesMotifs (exposeDesMotifsHtml: string): ExposeDesMotifs | null {
218
- const { document } = (new JSDOM(exposeDesMotifsHtml, {
219
- contentType: "text/html",
220
- })).window
221
- return transformExposeDesMotifs(document)
222
- }
223
-
224
- // Prevent from memory leak
225
- // https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
226
- export async function parseExposeDesMotifsFromFile (htmlFilePath: string): Promise<ExposeDesMotifs | null> {
227
- const { document } = (await JSDOM.fromFile(htmlFilePath, { contentType: "text/html" })).window
228
- return transformExposeDesMotifs(document)
229
- }