fmtr.tools 1.0.34__py3-none-any.whl → 1.0.35__py3-none-any.whl

fmtr/tools/pdf_tools.py

@@ -120,6 +120,32 @@ class Block(data_modelling_tools.Base):
         return self.bbox.rect
 
 
+class Page(data_modelling_tools.Base):
+    number: int
+    width: float
+    height: float
+    blocks: List[Block]
+
+    @property
+    def text(self) -> str:
+        """
+
+        Simple text representation
+
+        """
+        return ' '.join([block.text for block in self.blocks])
+
+    @classmethod
+    def from_dict(cls, data: Dict) -> Self:
+        """
+
+        Instantiate from PyMuPDF dictionary data
+
+        """
+
+        data['blocks'] = [Block.from_dict(block) for block in data['blocks']]
+        return cls(**data)
+
 class Document(pm.Document):
     """
 
@@ -128,21 +154,22 @@ class Document(pm.Document):
     """
 
     @property
-    def data(self) -> List[Block]:
+    def data(self) -> List[Page]:
         """
 
         Get representation of Document elements as Python objects.
 
         """
 
-        blocks = []
+        pages = []
 
-        for page in self:
-            for block in page.get_text("dict", flags=pm.TEXTFLAGS_TEXT | pm.TEXT_ACCURATE_BBOXES)["blocks"]:
-                obj = Block.from_dict(block)
-                blocks.append(obj)
+        for page_pm in self:
+            data = page_pm.get_text("dict", flags=pm.TEXTFLAGS_TEXT | pm.TEXT_ACCURATE_BBOXES)
+            data['number'] = page_pm.number
+            page = Page.from_dict(data)
+            pages.append(page)
 
-        return blocks
+        return pages
 
     def to_markdown(self, **kwargs) -> str:
         """

fmtr/tools/version

@@ -1 +1 @@
-1.0.34
+1.0.35

{fmtr.tools-1.0.34.dist-info → fmtr.tools-1.0.35.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: fmtr.tools
-Version: 1.0.34
+Version: 1.0.35
 Summary: Collection of high-level tools to simplify everyday development tasks, with a focus on AI/ML
 Home-page: https://github.com/fmtr/fmtr.tools
 Author: Frontmatter

{fmtr.tools-1.0.34.dist-info → fmtr.tools-1.0.35.dist-info}/RECORD

@@ -31,7 +31,7 @@ fmtr/tools/netrc_tools.py,sha256=PpNpz_mWlQi6VHGromKwFfTyLpHUXsd4LY6-OKLCbeI,376
 fmtr/tools/openai_tools.py,sha256=6SUgejgzUzmlKKct2_ePXntvMegu3FJgfk9x7aqtqYc,742
 fmtr/tools/parallel_tools.py,sha256=G__ZbLRRx4cP5OyqY1hKwnE-VI3m5prYABB0tnZHnes,3132
 fmtr/tools/path_tools.py,sha256=1GeWXdhV5rH99IfLI5ZFEnOJfs4Q4mYTT2R-rA791iQ,4273
-fmtr/tools/pdf_tools.py,sha256=mCLPJJlN2izIvSIjhJkEUZKT0GYQGmr96dPq9oIRsuo,3524
+fmtr/tools/pdf_tools.py,sha256=xvv9B84uAF81rFJRnXhSsxYuP42vY9ZdPVFrSMVe8G8,4069
 fmtr/tools/platform_tools.py,sha256=7p69CmAHe_sF68Fx9uVhns1k5EewTHTWgUYzkl6ZQKA,308
 fmtr/tools/process_tools.py,sha256=Ysh5Dk2QFBhXQerArjKdt7xZd3JrN5Ho02AaOjH0Nnw,1425
 fmtr/tools/profiling_tools.py,sha256=jpXVjaNKPydTasEQVNXvxzGtMhXPit08AnJddkU8uIc,46
@@ -42,7 +42,7 @@ fmtr/tools/string_tools.py,sha256=U2EptMWR6KDOP22ZQ4ReUHV4i25SP7xwCmZScI1sy4M,32
 fmtr/tools/tokenization_tools.py,sha256=9FP5vgPufWv0XA961eVKObFll0d_2mM0W3ut3rtZyeo,4329
 fmtr/tools/tools.py,sha256=xnfUrOnrT4OxFYez6vV5tAhydzCICJFiGVnviiZDEQo,796
 fmtr/tools/unicode_tools.py,sha256=yS_9wpu8ogNoiIL7s1G_8bETFFO_YQlo4LNPv1NLDeY,52
-fmtr/tools/version,sha256=95PJL7GqZ_Ze5aKBt6zAx5dswp7AauJVBg2AJ6DPaFU,6
+fmtr/tools/version,sha256=dl9h4ALGWOf6-smyLVw__eLZbZ3cQt7ygNYjZ5fNJZo,6
 fmtr/tools/version_tools.py,sha256=axzzHBS9V1n6YuSacsDKG3VfAvRqR8qr6aENCibR8vs,1248
 fmtr/tools/yaml_tools.py,sha256=Ol43ZwbnSXGnn1K98Uxx61KPGSqfC4axE-X2q1LKMwk,349
 fmtr/tools/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -53,9 +53,9 @@ fmtr/tools/tests/test_environment.py,sha256=iHaiMQfECYZPkPKwfuIZV9uHuWe3aE-p_dN_
 fmtr/tools/tests/test_json.py,sha256=IeSP4ziPvRcmS8kq7k9tHonC9rN5YYq9GSNT2ul6Msk,287
 fmtr/tools/tests/test_path.py,sha256=AkZQa6_8BQ-VaCyL_J-iKmdf2ZaM-xFYR37Kun3k4_g,2188
 fmtr/tools/tests/test_yaml.py,sha256=jc0TwwKu9eC0LvFGNMERdgBue591xwLxYXFbtsRwXVM,287
-fmtr.tools-1.0.34.dist-info/LICENSE,sha256=FW9aa6vVN5IjRQWLT43hs4_koYSmpcbIovlKeAJ0_cI,10757
-fmtr.tools-1.0.34.dist-info/METADATA,sha256=n17dBm5yzzI_ktSH_1hqUxN7cXJtCSRa_Fh4Vc_vqYQ,13438
-fmtr.tools-1.0.34.dist-info/WHEEL,sha256=oiQVh_5PnQM0E3gPdiz09WCNmwiHDMaGer_elqB3coM,92
-fmtr.tools-1.0.34.dist-info/entry_points.txt,sha256=CEStVkwJ1mTFvhN1WV5RdW83SkNW1d5Syj-KZ6A19ng,72
-fmtr.tools-1.0.34.dist-info/top_level.txt,sha256=t5341a8ii3n4RFizwTeXGmcq_pf4GqL1h9ylE5LIWRk,12
-fmtr.tools-1.0.34.dist-info/RECORD,,
+fmtr.tools-1.0.35.dist-info/LICENSE,sha256=FW9aa6vVN5IjRQWLT43hs4_koYSmpcbIovlKeAJ0_cI,10757
+fmtr.tools-1.0.35.dist-info/METADATA,sha256=oWxo_lXLyOF_Osu8JvqFeUTWS5lgh-aTRgGKYIzePkU,13438
+fmtr.tools-1.0.35.dist-info/WHEEL,sha256=oiQVh_5PnQM0E3gPdiz09WCNmwiHDMaGer_elqB3coM,92
+fmtr.tools-1.0.35.dist-info/entry_points.txt,sha256=CEStVkwJ1mTFvhN1WV5RdW83SkNW1d5Syj-KZ6A19ng,72
+fmtr.tools-1.0.35.dist-info/top_level.txt,sha256=t5341a8ii3n4RFizwTeXGmcq_pf4GqL1h9ylE5LIWRk,12
+fmtr.tools-1.0.35.dist-info/RECORD,,