[WIP] Working on the navigation parser

dapitains/app/ingest.py

@@ -1,7 +1,8 @@
-from typing import Dict, List, Optional, Any
+from typing import Dict, List, Optional, Any, Tuple
 from dapitains.app.database import Collection, Navigation, db
 from dapitains.metadata.xml_parser import Catalog
 from dapitains.tei.document import Document
+import copy
 
 def store_catalog(catalog: Catalog):
     for identifier, collection in catalog.objects.items():
@@ -23,17 +24,23 @@ def get_member_by_path(data: List[Dict[str, Any]], path: List[int]) -> Optional[
     """
     current_level = data
 
-    for index in path:
+    path_copy = [] + path
+    while path_copy:
+        index = path_copy.pop(0)
         try:
             current_level = current_level[index]
-            if 'members' in current_level:
+            if 'members' in current_level and path_copy:
                 current_level = current_level['members']
         except (IndexError, KeyError):
             return None
 
     return current_level
 
 
+def strip_members(obj: Dict[str, Any]) -> Dict[str, Any]:
+    return {k: v for k, v in obj.items() if k != "members"}
+
+
 def generate_paths(data: List[Dict[str, Any]], path: Optional[List[int]] = None) -> Dict[str, List[int]]:
     """
     Generate a dictionary mapping each 'ref' in a nested data structure to its path.
@@ -69,6 +76,54 @@ def recurse(items, current_path):
     recurse(data, [])
     return paths
 
+
+def get_nav(
+        refs: List[Dict[str, Any]],
+        paths: Dict[str, List[int]],
+        start_or_ref: Optional[str],
+        end: Optional[str],
+        down: Optional[int] = 1
+) -> Tuple[List[Dict[str, Any]], Optional[Dict[str, Any]], Optional[Dict[str, Any]]]:
+
+    paths_index = list(paths.keys())
+    start_index, end_index = None, None
+    if start_or_ref:
+        start_index = paths_index.index(start_or_ref)
+    if end:
+        end_index = paths_index.index(end) + 1
+
+    paths = dict(list(paths.items())[start_index:end_index])
+
+    current_level = [0]
+
+    start_path, end_path = None, None
+
+    if start_or_ref:
+        start_path = paths[start_or_ref]
+        current_level.append(len(start_path))
+    if end:
+        end_path = paths[end]
+        current_level.append(len(end_path))
+
+    current_level = max(current_level)
+
+    if down == -1:
+        down = max(list(map(len, paths.values())))
+
+    if down == 0:
+        paths = {key: value for key, value in paths.items() if len(value) == current_level}
+    else:
+        paths = {key: value for key, value in paths.items() if current_level < len(value) <= down + current_level}
+
+    return (
+        [
+            strip_members(get_member_by_path(refs, path)) for path in paths.values()
+        ],
+        strip_members(get_member_by_path(refs, start_path)) if start_path else None,
+        strip_members(get_member_by_path(refs, end_path)) if end_path else None
+    )
+
+
 if __name__ == "__main__":
     import flask
     import os

tests/test_db_create.py

@@ -1,5 +1,5 @@
 import flask
-from dapitains.app.ingest import generate_paths, get_member_by_path
+from dapitains.app.ingest import generate_paths, get_member_by_path, get_nav, strip_members
 from dapitains.tei.document import Document
 import os
 
@@ -48,5 +48,30 @@ def test_simple_path():
             "Mark 1:3": [1, 0, 3]
         }
     }
+    assert strip_members(get_member_by_path(refs[None], paths[None]["Luke"])) == {'citeType': 'book', 'ref': 'Luke'}
     assert get_member_by_path(refs[None], paths[None]["Mark 1:3"]) == {'citeType': 'verse', 'ref': 'Mark 1:3'}
 
+
+def test_navigation():
+    doc = Document(f"{local_dir}/tei/base_tei.xml")
+    refs = {
+        tree: [ref.json() for ref in obj.find_refs(doc.xml, structure=obj.units)]
+        for tree, obj in doc.citeStructure.items()
+    }
+    paths = {tree: generate_paths(ref) for tree, ref in refs.items()}
+    assert get_nav(refs[None], paths[None], start_or_ref=None, end=None, down=1) == ([
+        {'citeType': 'book', 'ref': 'Luke'},
+        {'citeType': 'book', 'ref': 'Mark'}
+    ], None, None)
+    assert get_nav(refs[None], paths[None], start_or_ref="Luke 1:1", end="Luke 1#1", down=0) == ([
+        {'citeType': 'verse', 'ref': 'Luke 1:1'},
+        {'citeType': 'verse', 'ref': 'Luke 1:2'},
+        {'citeType': 'bloup', 'ref': 'Luke 1#1'}
+    ], {'citeType': 'verse', 'ref': 'Luke 1:1'}, {'citeType': 'bloup', 'ref': 'Luke 1#1'})
+    assert get_nav(refs[None], paths[None], start_or_ref="Luke 1:1", end="Mark 1:2", down=0) == ([
+        {'citeType': 'verse', 'ref': 'Luke 1:1'},
+        {'citeType': 'verse', 'ref': 'Luke 1:2'},
+        {'citeType': 'bloup', 'ref': 'Luke 1#1'},
+        {'citeType': 'verse', 'ref': 'Mark 1:1'},
+        {'citeType': 'verse', 'ref': 'Mark 1:2'}
+    ], {'citeType': 'verse', 'ref': 'Luke 1:1'}, {'citeType': 'verse', 'ref': 'Mark 1:2'})