datamule 1.4.0__tar.gz → 1.4.3__tar.gz

{datamule-1.4.0 → datamule-1.4.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: datamule
-Version: 1.4.0
+Version: 1.4.3
 Summary: Work with SEC submissions at scale.
 Home-page: https://github.com/john-friedman/datamule-python
 Author: John Friedman

{datamule-1.4.0 → datamule-1.4.3}/datamule/document/document.py

@@ -3,10 +3,10 @@ import csv
 import re
 from doc2dict import xml2dict, txt2dict, dict2dict
 from doc2dict.mapping import flatten_hierarchy
-from doc2dict import html2dict, visualize_dict, get_title, unnest_dict
+from doc2dict import html2dict, visualize_dict, get_title, unnest_dict, pdf2dict
 from ..mapping_dicts.txt_mapping_dicts import dict_10k, dict_10q, dict_8k, dict_13d, dict_13g
 from ..mapping_dicts.xml_mapping_dicts import dict_345
-from ..mapping_dicts.html_mapping_dicts import dict_10k_html, dict_10q_html, dict_8k_html
+from ..mapping_dicts.html_mapping_dicts import *
 from selectolax.parser import HTMLParser
 from .processing import process_tabular_data
 from pathlib import Path
@@ -120,12 +120,31 @@ class Document:
             self.data = {}
             self.data['document'] = dict2dict(txt2dict(content=content, mapping_dict=mapping_dict))
         elif self.extension in ['.htm', '.html']:
-            if self.type == '10-K':
+            
+            if self.type == '1-K':
+                mapping_dict = dict_1kpartii_html
+            elif self.type == '1-SA':
+                mapping_dict = dict_1sa_html
+            elif self.type == '1-U':
+                mapping_dict = dict_1u_html
+            elif self.type == '10-12B':
+                mapping_dict = dict_1012b_html
+            elif self.type == '10-D':
+                mapping_dict = dict_10d_html
+            elif self.type == '10-K':
                 mapping_dict = dict_10k_html
             elif self.type == '10-Q':
                 mapping_dict = dict_10q_html
+            elif self.type == '20-F':
+                mapping_dict = dict_20f_html
             elif self.type == '8-K':
                 mapping_dict = dict_8k_html
+            elif self.type == 'ABS-15G':
+                mapping_dict = dict_abs15g_html
+            elif self.type == 'SD':
+                mapping_dict = dict_sd_html
+            elif self.type in ['NT 10-K', 'NT 10-Q','NT 20-F']:
+                mapping_dict = dict_nt10k_html
             
             dct = html2dict(content=self.content, mapping_dict=mapping_dict)
             self.data = dct
@@ -134,6 +153,8 @@ class Document:
                 mapping_dict = dict_345
             
             self.data = xml2dict(content=self.content, mapping_dict=mapping_dict)
+        elif self.extension == '.pdf':
+            self.data = pdf2dict(content=self.content, mapping_dict=mapping_dict)
         else:
             pass
     
@@ -235,8 +256,7 @@ class Document:
         return result
 
    
-   # this will all have to be changed. default will be to flatten everything
-   # candidate for deletion
+   # TODO CHANGE THIS
     def __iter__(self):
         self.parse()
 

datamule-1.4.3/datamule/mapping_dicts/html_mapping_dicts.py

@@ -0,0 +1,48 @@
+dict_10k_html = {
+    ('part',r'^part\s*([ivx]+)$') : 0,
+    ('signatures',r'^signatures?\.*$') : 0,
+    ('item',r'^item\s*(\d+)\.?([a-z])?') : 1,
+}
+dict_10q_html = dict_10k_html
+
+dict_8k_html = {
+    ('signatures',r'^signatures?\.*$') : 0,
+    ('item',r'^item\s*(\d+\.\d+)') : 0,
+}
+
+dict_sd_html = {
+    ('signatures',r'^signatures?\.*$') : 0,
+    ('item',r'^item\s*(\d+\.\d+)') : 0,
+}
+
+dict_abs15g_html = {
+    ('part',r'^part\s*([ivx]+)') : 0,
+    ('signatures',r'^signatures?\.*$') : 0,
+    ('item',r'^item\s*(\d+\.\d+)') : 1,
+}
+
+
+
+dict_nt10k_html = {
+    ('part',r'^part\s*([ivx]+)') : 0,
+}
+
+dict_1kpartii_html = {
+    ('item',r'^item\s*(\d+)') : 1,
+}
+
+dict_1sa_html = dict_1kpartii_html
+
+dict_1u_html = {('item',r'^item\s*(\d+)') : 1,
+                ('signatures',r'^signatures?\.*$') : 1,}
+
+dict_1012b_html = dict_1u_html
+
+dict_10d_html = dict_10k_html
+
+dict_20f_html = {
+    ('part',r'^part\s*([ivx]+)') : 0,
+    ('item',r'^item\s*(\d+)\.?([a-z])?') : 1,
+    ('letter',r'\d*\.?([a-z])') : 2,
+    ('signatures',r'^signatures?\.*$') : 0,
+}

{datamule-1.4.0 → datamule-1.4.3}/datamule.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: datamule
-Version: 1.4.0
+Version: 1.4.3
 Summary: Work with SEC submissions at scale.
 Home-page: https://github.com/john-friedman/datamule-python
 Author: John Friedman

{datamule-1.4.0 → datamule-1.4.3}/setup.py

@@ -32,7 +32,7 @@ if not os.path.exists(file_path):
 setup(
     name="datamule",
     author="John Friedman",
-    version="1.4.0",
+    version="1.4.3",
     description="Work with SEC submissions at scale.",
     packages=find_packages(include=['datamule', 'datamule.*']),
     url="https://github.com/john-friedman/datamule-python",

datamule-1.4.0/datamule/mapping_dicts/html_mapping_dicts.py

@@ -1,11 +0,0 @@
-dict_10k_html = {
-    ('part',r'^part\s*([ivx]+)$') : 0,
-    ('signatures',r'^signatures?\.*$') : 0,
-    ('item',r'^item\s*(\d+)\.?([a-z])?') : 1,
-}
-dict_10q_html = dict_10k_html
-
-dict_8k_html = {
-    ('signatures',r'^signatures?\.*$') : 0,
-    ('item',r'^item\s*(\d+\.\d+)') : 0,
-}