feat: enhance ARRAYData initialization and transform method with signature handling and data extraction improvements

src/transformers/transformers.py

@@ -117,7 +117,8 @@ class BaseStruct:
 
 
 class ARRAYData:
-    def __init__(self):
+    def __init__(self, signature=None) -> None:
+        self.signature = None
         pass
 
     def transform(self, data: str) -> BookData:
@@ -216,20 +217,72 @@ class ARRAYData:
                 titledata = data.split("[ti_long_f]")[1].split(")\n")[0].strip()
                 title = titledata.split("=>")[1].strip().split("/")[0].strip()
             return title
+
+        def _get_adis_idn(data, signature):
+            loksatz_match = re.search(
+                r"\[loksatz\] => Array\s*\((.*?)\)", data, re.DOTALL
+            )
+            if loksatz_match:
+                loksatz_content = loksatz_match.group(1)
+
+                # Step 2: Extract JSON objects within the loksatz section
+                json_objects = re.findall(r"{.*?}", loksatz_content, re.DOTALL)
+                # Print each JSON object
+                for obj in json_objects:
+                    data = eval(obj)
+                    if data["signatur"] == signature:
+                        return data["adis_idn"]
+
+        def _get_in_apparat(data):
+            loksatz_match = re.search(
+                r"\[loksatz\] => Array\s*\((.*?)\)", data, re.DOTALL
+            )
+            if loksatz_match:
+                loksatz_content = loksatz_match.group(1)
+
+                # Step 2: Extract JSON objects within the loksatz section
+                json_objects = re.findall(r"{.*?}", loksatz_content, re.DOTALL)
+                # Print each JSON object
+                for obj in json_objects:
+                    data = eval(obj)
+                    if data["ausleihcode"] == "R" and data["standort"] == "40":
+                        return True
+                    else:
+                        return False
+
+        ppn = _get_line(data, "[kid]")
+        title = _get_title(data).strip()
+        author = _get_author(data)
+        edition = _get_list_entry(data, "[ausgabe]", "[0]").replace(",", "")
+        link = f"https://rds.ibs-bw.de/phfreiburg/link?kid={_get_line(data,'[kid]')}"
+        isbn = _get_isbn(data)
+        # [self._get_list_entry(data,"[isbn]","[0]"),self._get_list_entry(data,"[is]","[1]")],
+        language = _get_list_entry(data, "[la_facet]", "[0]")
+        publisher = _get_list_entry(data, "[pu]", "[0]")
+        year = _get_list_entry(data, "[py_display]", "[0]")
+        pages = _get_list_entry(data, "[umfang]", "[0]").split(":")[0].strip()
+        signature = (
+            self.signature if self.signature is not None else _get_signature(data)
+        )
+
+        place = _get_list_entry(data, "[pp]", "[0]")
+        adis_idn = _get_adis_idn(data, signature=signature)
+        in_apparat = _get_in_apparat(data)
         return BookData(
-            ppn=_get_line(data, "[kid]"),
+            ppn=ppn,
-            title=_get_title(data).strip(),
+            title=title,
-            author=_get_author(data),
+            author=author,
-            edition=_get_list_entry(data, "[ausgabe]", "[0]").replace(",", ""),
+            edition=edition,
-            link=f"https://rds.ibs-bw.de/phfreiburg/link?kid={_get_line(data,'[kid]')}",
+            link=link,
-            isbn=_get_isbn(data),
+            isbn=isbn,
-            # [self._get_list_entry(data,"[isbn]","[0]"),self._get_list_entry(data,"[is]","[1]")],
+            language=language,
-            language=_get_list_entry(data, "[la_facet]", "[0]"),
+            publisher=publisher,
-            publisher=_get_list_entry(data, "[pu]", "[0]"),
+            year=year,
-            year=_get_list_entry(data, "[py_display]", "[0]"),
+            pages=pages,
-            pages=_get_list_entry(data, "[umfang]", "[0]").split(":")[0].strip(),
+            signature=signature,
-            signature=_get_signature(data),
+            place=place,
-            place=_get_list_entry(data, "[pp]", "[0]"),
+            adis_idn=adis_idn,
+            in_apparat=in_apparat,
         )