From 5a57bf4d32b34237ab5578e89acf999a838fe48d Mon Sep 17 00:00:00 2001
From: Kiryl <kiryl.miatselitsa@teqniksoft.com>
Date: Thu, 13 Oct 2022 11:05:32 +0300
Subject: [PATCH] docx presets improves

---
 presets/docx_presets.json         | 86 +++++++++++++++++++++++++------
 src/docx_converter/docx_solver.py |  4 +-
 2 files changed, 72 insertions(+), 18 deletions(-)

diff --git a/presets/docx_presets.json b/presets/docx_presets.json
index fed2d62..96f861b 100644
--- a/presets/docx_presets.json
+++ b/presets/docx_presets.json
@@ -12,9 +12,13 @@
                             "name": "id",
                             "value": "^Table of Contents\\d+"
                         }
-                    ]
+                    ],
+                    "text": null
                 },
-                "tag_to_wrap": "TOC"
+                "tag_to_wrap": {
+                    "name": "TOC",
+                    "attrs": []
+                }
             }
         ]
     },
@@ -35,7 +39,8 @@
                             "name": "id",
                             "value": "^Table of Contents\\d+"
                         }
-                    ]
+                    ],
+                    "text": null
                 }
             }
         ]
@@ -46,7 +51,10 @@
             {
                 "tags": ["^h[6-9]$"],
                 "condition": null,
-                "tag_to_replace": "p"
+                "tag_to_replace": {
+                    "name": "p",
+                    "attrs": null
+                }
             },
             {
                 "tags": ["^div$"],
@@ -58,9 +66,13 @@
                             "name": "style",
                             "value": "column-count: 2"
                         }
-                    ]
+                    ],
+                    "text": null
                 },
-                "tag_to_replace": "p"
+                "tag_to_replace": {
+                    "name": "p",
+                    "attrs": null
+                }
             }
         ]
     },
@@ -68,12 +80,14 @@
         "preset_name": "attr_replacer",
         "rules": [
             {
-                "attr": {
-                    "name": "style",
-                    "value": "column-count: 2"
-                },
+                "tags": ["^p$"],
                 "condition": {
-                    "tags": ["^p$"]
+                    "attrs": [
+                        {
+                            "name": "style",
+                            "value": "column-count: 2"
+                        }
+                    ]
                 },
                 "attr_to_replace": {
                     "name": "class",
@@ -103,7 +117,8 @@
                             "name": "face",
                             "value": "^Times New Roman[\\w, ]+$"
                         }
-                    ]
+                    ],
+                    "text": null
                 }
             },
             {
@@ -111,7 +126,8 @@
                 "condition": {
                     "parent_tags": ":is(li)",
                     "child_tags": null,
-                    "attrs": null
+                    "attrs": null,
+                    "text": null
                 }
             },
             {
@@ -124,7 +140,8 @@
                             "name": "name",
                             "value": "_GoBack"
                         }
-                    ]
+                    ],
+                    "text": null
                 }
             },
             {
@@ -132,7 +149,8 @@
                 "condition": {
                     "parent_tags": ":is(a)",
                     "child_tags": ":is(a)",
-                    "attrs": null
+                    "attrs": null,
+                    "text": null
                 }
             },
             {
@@ -140,7 +158,8 @@
                 "condition": {
                     "parent_tags": ":is(h1, h2, h3, h4, h5, h6, h7, h8, h9)",
                     "child_tags": null,
-                    "attrs": null
+                    "attrs": null,
+                    "text": null
                 }
             },
             {
@@ -148,5 +167,40 @@
                 "condition": null
             }
         ]
+    },
+    {
+        "preset_name": "inserter",
+        "rules": [
+            {
+                "tags": ["^p$"],
+                "condition": {
+                    "parent_tags": null,
+                    "child_tags": null,
+                    "attrs": null,
+                    "text": "\\$\\$[\\s\\S]*?\\$\\$"
+                },
+                "tag_to_insert": {
+                    "name": "span",
+                    "attrs": [
+                        {
+                            "name": "class",
+                            "value": "math-tex"
+                        }
+                    ]
+                }
+            }
+        ]
+    },
+    {
+        "preset_name": "text_replacer",
+        "rules": [
+            {
+                "tags": ["^p$"],
+                "condition": {
+                    "text": "(\\\\nonumber\\\\\\\\\\\\noalign{\\\\pagebreak}[\\s\\S]*?)\\\\"
+                },
+                "text_to_replace": "\\\\"
+            }
+        ]
     }
 ]
diff --git a/src/docx_converter/docx_solver.py b/src/docx_converter/docx_solver.py
index 3cd324d..abb13e3 100644
--- a/src/docx_converter/docx_solver.py
+++ b/src/docx_converter/docx_solver.py
@@ -80,7 +80,8 @@ class DocxBook(BookSolver):
 
 
 if __name__ == "__main__":
-    docx_file_path = "../../books/docx/AmericanGovernment3e-WEB.docx"
+
+    docx_file_path = "../../books/docx/output.docx"
     logger_object = BookLogger(
         name="docx", book_id=docx_file_path.split("/")[-1])
     locker = Event()
@@ -88,7 +89,6 @@ if __name__ == "__main__":
 
     html_converter = Docx2LibreHtml(file_path=docx_file_path,
                                     logger=logger_object, libre_locker=locker)
-
     html_preprocessor = HtmlPresetsProcessor(
         logger=logger_object, preset_path="../../presets/docx_presets.json")
     style_preprocessor = StyleReader()