Further work on table referencing and basic code refactor

setup.py

@@ -1,3 +1,3 @@
 from setuptools import setup
 
-setup(version="0.0.2")
+setup(version="0.0.3")

src/paradoc/common.py

@@ -27,6 +27,7 @@ class Table:
     df: pd.DataFrame
     caption: str
     format: TableFormat = TableFormat()
+    add_link: bool = False
 
     def to_markdown(self, include_name_in_cell=False, flags=None):
         df = self.df.copy()
@@ -36,7 +37,9 @@ def to_markdown(self, include_name_in_cell=False, flags=None):
         tbl_str = df.to_markdown(index=False, tablefmt="grid")
         if flags is not None and TableFlags.NO_CAPTION in flags:
             return tbl_str
-        tbl_str += f"\nTable: {self.caption} {{#tbl:{self.name}}}"
+        tbl_str += f"\n\nTable: {self.caption}"
+        if self.add_link:
+            tbl_str += f"{{#tbl:{self.name}}}"
         return tbl_str
 
 
@@ -70,10 +73,9 @@ def to_latex(self, print_latex=False, print_formula=False, flags=None):
 
 
 @dataclass
-class Formatting:
-    is_appendix: bool
-    paragraph_style_map: dict
-    table_format: TableFormat
+class DocXFormat:
+    pg_font: str = "Arial"
+    pg_size: int = 11
 
 
 @dataclass

src/paradoc/document.py

@@ -9,7 +9,14 @@
 import pandas as pd
 import pypandoc
 
-from .common import Equation, ExportFormats, MarkDownFile, Table, TableFormat
+from .common import (
+    DocXFormat,
+    Equation,
+    ExportFormats,
+    MarkDownFile,
+    Table,
+    TableFormat,
+)
 from .utils import get_list_of_files, variable_sub
 
 
@@ -61,9 +68,9 @@ def __init__(
         self.variables = dict()
         self.tables: Dict[str, Table] = dict()
         self.equations: Dict[str, Equation] = dict()
+        self.doc_format = DocXFormat()
 
         # Style info: https://python-docx.readthedocs.io/en/latest/user/styles-using.html
-        self.table_format = TableFormat()
         self.paragraph_style_map = kwargs.get("paragraph_style_map", OneDoc.default_paragraph_map)
         self.appendix_heading_map = kwargs.get("appendix_heading_map", OneDoc.default_app_map)
 

src/paradoc/io/word/common.py

@@ -6,10 +6,11 @@
 from docx.shared import Pt
 from docx.table import Table as DocxTable
 from docx.text.paragraph import Paragraph
+from docx.text.run import Run
 
 from paradoc.common import Table
 
-from .references import insert_caption_into_runs
+from .references import add_seq_reference
 
 
 @dataclass
@@ -25,7 +26,6 @@ def is_complete(self):
         return all([x is not None for x in docx_attr])
 
     def format_table(self, is_appendix):
-
         tbl = self.docx_table
         tbl_format = self.table_ref.format
 
@@ -45,15 +45,50 @@ def format_table(self, is_appendix):
                             font.bold = True
                         else:
                             font.bold = False
+        tbl.autofit = True
 
         # Format table Caption
         caption = self.docx_caption
         caption.paragraph_format.alignment = WD_PARAGRAPH_ALIGNMENT.CENTER
-        insert_caption_into_runs(caption, "Table", is_appendix)
+
+        rebuild_caption(caption, self.table_ref.caption, is_appendix)
+
         for run in caption.runs:
             run.font.name = tbl_format.font_style
 
         # Fix formatting after Table
-        follower_pg = self.docx_following_pg
-        follower_pg.runs[0].text = "\n" + follower_pg.runs[0].text
-        follower_pg.paragraph_format.space_before = None
+
+        self.docx_following_pg.paragraph_format.space_before = Pt(12)
+        # follower_pg = self.docx_following_pg
+        #
+        # i = par_index(follower_pg)
+        #
+        # follower_pg.runs[0].text = "\n" + follower_pg.runs[0].text
+        # follower_pg.paragraph_format.space_before = None
+
+
+def rebuild_caption(caption: Paragraph, caption_str, is_appendix):
+    caption.clear()
+    caption.runs.clear()
+
+    run = caption.add_run()
+
+    heading_ref = "Appendix" if is_appendix is True else '"Heading 1"'
+
+    seq1 = caption._element._new_r()
+    seq1.text = "Table "
+
+    add_seq_reference(seq1, f"STYLEREF \\s {heading_ref} \\n", run._parent)
+    run._element.addprevious(seq1)
+
+    stroke = caption._element._new_r()
+    new_run = Run(stroke, run._parent)
+    new_run.text = "-"
+    run._element.addprevious(stroke)
+    seq2 = caption._element._new_r()
+    add_seq_reference(seq2, "SEQ Table \\* ARABIC \\s 1", run._parent)
+    run._element.addprevious(seq2)
+    fin = caption._element._new_r()
+    fin_run = Run(fin, run._parent)
+    fin_run.text = ": " + caption_str
+    run._element.addprevious(fin)

src/paradoc/io/word/exporter.py

@@ -4,6 +4,7 @@
 import re
 from typing import List, Union
 
+import numpy as np
 from docx import Document
 from docx.table import Table as DocxTable
 from docx.text.paragraph import Paragraph
@@ -32,6 +33,7 @@ def convert_to_docx(self, output_name, dest_file):
         composer_app = add_to_composer(MY_DOCX_TMPL_BLANK, one_doc.md_files_app)
 
         for tbl in self.identify_tables(composer_main.doc):
+
             tbl.format_table(is_appendix=False)
 
         for tbl in self.identify_tables(composer_app.doc):
@@ -71,12 +73,12 @@ def identify_tables(self, doc: Document):
                 continue
 
             if block.style.name == "Table Caption":
+                if "using solid elements" in block.text:
+                    print("sd")
                 current_table.docx_caption = block
 
-            if type(block) == Paragraph and prev_table is True and len(block.runs) > 0:
-                block.runs[0].text = "\n" + block.runs[0].text
+            if type(block) == Paragraph and prev_table is True:
                 prev_table = False
-                block.paragraph_format.space_before = None
                 current_table.docx_following_pg = block
 
             if current_table.is_complete():
@@ -90,19 +92,42 @@ def identify_tables(self, doc: Document):
 
         return tables
 
-    def get_related_table(self, current_table: DocXTableRef) -> Union[Table, None]:
+    def get_related_table(self, current_table: DocXTableRef, frac=1e-4) -> Union[Table, None]:
         one = self.one_doc
+
+        # Search using Caption string
         caption = current_table.docx_caption
-        re_cap = re.compile("Table [0-9]{0,9}:(.*)")
+        re_cap = re.compile(r"Table\s*[0-9]{0,9}:(.*)")
         for key, tbl in one.tables.items():
             if "Table" in caption.text:
                 m = re_cap.search(caption.text)
+                if m is None:
+                    raise ValueError()
                 caption_text = str(m.group(1).strip())
             else:
                 caption_text = str(caption.text)
             caption_text = caption_text.replace("”", '"')
             if tbl.caption == caption_text:
                 return tbl
+
+        # If no match using caption string, then use contents of table
+        content = get_first_row_from_table(current_table.docx_table)
+        is_content_numeric = False
+
+        try:
+            content_numeric = np.array(content, dtype=float)
+            is_content_numeric = True
+        except ValueError:
+            content_numeric = None
+
+        for key, tbl in one.tables.items():
+            row_1 = tbl.df.iloc[0].values
+            if is_content_numeric and len(content) == len(row_1):
+                tot = sum(row_1)
+                diff = sum(row_1 - content_numeric)
+                if abs(diff) < abs(tot) * frac:
+                    return tbl
+            print("")
         return None
 
 
@@ -116,3 +141,15 @@ def add_to_composer(source_doc, md_files: List[MarkDownFile]) -> Composer:
         composer_doc.append(doc_in)
         logging.info(f"Added {md.new_file}")
     return composer_doc
+
+
+def get_first_row_from_table(docx_table: DocxTable, num_row=1):
+    content = []
+    for i, row in enumerate(docx_table.rows):
+        if i == 0:
+            continue
+        for cell in row.cells:
+            paragraphs = cell.paragraphs
+            for paragraph in paragraphs:
+                content.append(paragraph.text.strip())
+        return content

src/paradoc/io/word/references.py

@@ -143,3 +143,19 @@ def add_seq_reference(run_in, seq, parent):
     fldChar.set(qn("w:fldCharType"), "end")
     r.append(fldChar)
     return new_run
+
+
+def add_table_reference(paragraph, seq=" SEQ Table \\* ARABIC \\s 1"):
+    run = paragraph.add_run()
+    r = run._r
+    fldChar = OxmlElement("w:fldChar")
+    fldChar.set(qn("w:fldCharType"), "begin")
+    r.append(fldChar)
+    instrText = OxmlElement("w:instrText")
+    instrText.text = seq
+    r.append(instrText)
+    fldChar = OxmlElement("w:fldChar")
+    fldChar.set(qn("w:fldCharType"), "end")
+    r.append(fldChar)
+
+    return run