Better legifrance inclusion

compiler/literate/latex.ml

@@ -151,7 +151,8 @@ codes={\catcode`\$=3\catcode`\^=7}
 \tableofcontents
 
 \[\star\star\star\]
-\clearpage|latex}
+\clearpage
+|latex}
     (match language with Fr -> "french" | En -> "english" | Pl -> "polish")
     (match language with Fr -> "\\setmainfont{Marianne}" | _ -> "")
     (* for France, we use the official font of the French state design system

compiler/surface/parser.mly

@@ -662,7 +662,7 @@ let source_file_item :=
   END_DIRECTIVE ; {
   let filename = String.trim (String.concat "" args) in
   let pos = Pos.from_lpos $sloc in
-  let jorftext = Re.Pcre.regexp "JORFTEXT\\d{12}" in
+  let jorftext = Re.Pcre.regexp "(JORFARTI\\d{12}|LEGIARTI\\d{12}|CETATEXT\\d{12})" in
   if Re.Pcre.pmatch ~rex:jorftext filename && page = None then
     LawInclude (Ast.LegislativeText (filename, pos))
   else if Filename.extension filename = ".pdf" || page <> None then

french_law/catala_legifrance/api.ml

@@ -191,13 +191,6 @@ let raise_article_parsing_error
     (Yojson.Basic.to_string json);
   exit 1
 
-type law_excerpt = Yojson.Basic.t
-
-let retrieve_law_excerpt (access_token : string) (text_id : string) :
-    law_excerpt =
-  run_request
-    (make_request access_token "consult/jorfPart" ["textCid", text_id])
-
 let get_article_id (article : article) : string =
   try
     article.content
@@ -237,6 +230,18 @@ let get_article_text (article : article) : string =
   with Yojson.Basic.Util.Type_error (msg, obj) ->
     raise_article_parsing_error article.content msg obj
 
+let get_article_title (article : article) : string =
+  try
+    article.content
+    |> Yojson.Basic.Util.member
+         (match article.typ with
+         | CETATEXT -> "text"
+         | LEGIARTI | JORFARTI -> "article")
+    |> Yojson.Basic.Util.member "titre"
+    |> Yojson.Basic.Util.to_string
+  with Yojson.Basic.Util.Type_error (msg, obj) ->
+    raise_article_parsing_error article.content msg obj
+
 let get_article_expiration_date (article : article) : Unix.tm =
   try
     let article_id = get_article_id article in
@@ -284,50 +289,3 @@ let get_article_new_version (article : article) : string =
       |> Yojson.Basic.Util.to_string
     with Yojson.Basic.Util.Type_error (msg, obj) ->
       raise_article_parsing_error article.content msg obj)
-
-let get_law_excerpt_title (json : law_excerpt) : string =
-  json |> Yojson.Basic.Util.member "title" |> Yojson.Basic.Util.to_string
-
-type law_excerpt_article = { id : string; num : string; content : string }
-
-let clean_html (s : string) : string =
-  let new_line = Re.Pcre.regexp "\\s*\\<br\\s*\\/\\>\\s*" in
-  let s = Re.Pcre.substitute ~rex:new_line ~subst:(fun _ -> "\n") s in
-  let tag = Re.Pcre.regexp "\\<[^\\>]+\\>" in
-  let s = Re.Pcre.substitute ~rex:tag ~subst:(fun _ -> "") s in
-  String.trim s
-
-let get_law_excerpt_articles (json : law_excerpt) : law_excerpt_article list =
-  let articles =
-    json |> Yojson.Basic.Util.member "articles" |> Yojson.Basic.Util.to_list
-  in
-  let articles =
-    List.sort
-      (fun a1 a2 ->
-        let a1_num =
-          int_of_string
-            (a1 |> Yojson.Basic.Util.member "num" |> Yojson.Basic.Util.to_string)
-        in
-        let a2_num =
-          int_of_string
-            (a2 |> Yojson.Basic.Util.member "num" |> Yojson.Basic.Util.to_string)
-        in
-        compare a1_num a2_num)
-      articles
-  in
-  List.map
-    (fun article ->
-      let article_id =
-        article |> Yojson.Basic.Util.member "id" |> Yojson.Basic.Util.to_string
-      in
-      let article_num =
-        article |> Yojson.Basic.Util.member "num" |> Yojson.Basic.Util.to_string
-      in
-      let article_content =
-        article
-        |> Yojson.Basic.Util.member "content"
-        |> Yojson.Basic.Util.to_string
-        |> clean_html
-      in
-      { id = article_id; num = article_num; content = article_content })
-    articles

french_law/catala_legifrance/api.mli

@@ -42,26 +42,12 @@ val retrieve_article : access_token -> article_id -> article
 (** [retrieve_article token article_id] returns the article from the LegiFrance
     API.*)
 
-type law_excerpt
-
-val retrieve_law_excerpt : access_token -> string -> law_excerpt
-(**[retrieve_law_excerpt token excerpt_id] returns a whole excerpt of a
-   legislative statute from the LegiFrance API. [excerpt_id] should be of the
-   form ["JORFTEXT000033736934"] *)
-
 (**{2 Manipulating API objects}*)
 
 (**{3 Articles}*)
 
 val get_article_id : article -> string
 val get_article_text : article -> string
+val get_article_title : article -> string
 val get_article_expiration_date : article -> Unix.tm
 val get_article_new_version : article -> string
-
-(**{3 Law excerpts}*)
-
-val get_law_excerpt_title : law_excerpt -> string
-
-type law_excerpt_article = { id : string; num : string; content : string }
-
-val get_law_excerpt_articles : law_excerpt -> law_excerpt_article list

french_law/catala_legifrance/catala_legifrance.ml

@@ -165,25 +165,14 @@ let compare_to_versions
 let include_legislative_text
     (id : string * Pos.t)
     (access_token : Api.access_token) : string =
-  let excerpt = Api.retrieve_law_excerpt access_token (fst id) in
-  let title = "#" ^ Api.get_law_excerpt_title excerpt in
-  let excerpts = Api.get_law_excerpt_articles excerpt in
-  let text_to_return =
-    String.concat "\n\n"
-      (List.map (fun article -> article.Api.content) excerpts)
-  in
-  let articles =
-    List.map
-      (fun article ->
-        Printf.sprintf "## Article %s|%s@\n%s" article.Api.num article.Api.id
-          article.Api.content)
-      excerpts
-  in
-  let to_insert = title ^ "\n\n" ^ String.concat "\n\n" articles in
   let pos = snd id in
+  let id = Api.parse_id (fst id) in
+  let article = Api.retrieve_article access_token id in
+  let text_to_return = Api.get_article_text article in
+  let to_insert = text_to_return in
   Cli.debug_format "Position: %s" (Pos.to_string_short pos);
   let file = Pos.get_file pos in
-  let include_line = Pos.get_end_line pos in
+  let include_line = Pos.get_start_line pos in
   let ic = open_in file in
   let new_file = file ^ ".new" in
   Cli.warning_print "LégiFrance inclusion detected, writing new contents to %s"