pyeasyphd 0.0.2__py3-none-any.whl

pyeasyphd/tools/spider/process_spider_url.py

@@ -0,0 +1,74 @@
+import os
+import re
+from typing import Dict, List
+
+from pyadvtools import IterateSortDict, read_list, sort_int_str, write_list
+
+
+class ProcessSpiderUrl(object):
+    """Process spider URL.
+
+    Args:
+        path_abbr (str): path to abbreviation
+        abbr_standard (str): abbreviation standard
+
+    Attributes:
+        path_abbr (str): path to abbreviation
+        abbr_standard (str): abbreviation standard
+    """
+
+    def __init__(self, path_abbr, abbr_standard: str) -> None:
+        self.path_abbr = os.path.expandvars(os.path.expanduser(path_abbr))
+        self.abbr_standard = abbr_standard
+
+    def check_spider_url(
+        self,
+        folder_start_swith: str = "url",
+        extension: str = ".txt",
+        write_flag: bool = False,
+        iterate_check_url: bool = False,
+    ) -> None:
+        if os.path.exists(self.path_abbr):
+            for i in [f for f in os.listdir(self.path_abbr) if f.startswith(folder_start_swith)]:
+                print("*" * 5 + f" Check *{extension} ./{os.path.basename(self.path_abbr)}/{i}")
+                self._check_delete(os.path.join(self.path_abbr, i), extension, write_flag, iterate_check_url)
+
+    def _check_delete(
+        self, path_storage: str, extension: str = ".txt", write_flag: bool = False, iterate_check_url: bool = False
+    ) -> None:
+        data_dict: Dict[str, Dict[str, List[str]]] = {}
+        files = [f for f in os.listdir(path_storage) if f.endswith(extension)]
+        for f in files:
+            mch = re.match(r"([a-zA-Z]+)_([\w\-]+)", f)
+            if mch:
+                a, b = mch.groups()
+                data_dict.setdefault(a, {}).setdefault(b, []).extend(read_list(f, "r", path_storage))
+
+        for a in IterateSortDict(False).dict_update(data_dict):
+            b_list = sort_int_str(list(data_dict[a].keys()))
+
+            for b in b_list:
+                new_temp_list: List[str] = []
+                duplicate_dict: Dict[str, List[str]] = {}
+                for line in data_dict[a][b]:
+                    line_flag = False
+                    if iterate_check_url:
+                        for bb in b_list[: b_list.index(b)]:
+                            if line in data_dict[a][bb]:
+                                duplicate_dict.setdefault(f"{a}_{b} and {a}_{bb}", []).append(line.strip())
+                                line_flag = True
+                                break
+
+                    if (not line_flag) and (line not in new_temp_list):
+                        new_temp_list.append(line)
+                    elif line in new_temp_list:
+                        print(f"Duplicate item {line.strip()} in {a}_{b}{extension}.\n")
+
+                data_dict[a][b] = new_temp_list
+
+                if duplicate_dict:
+                    print(f"Duplicate items in {extension}:\n", duplicate_dict)
+
+                if write_flag:
+                    write_list(new_temp_list, f"{a}_{b}{extension}", "w", path_storage, False, True, True, True)
+        return None

pyeasyphd/tools/spider/process_spider_url_bib.py

@@ -0,0 +1,62 @@
+import os
+from typing import Any, Dict
+
+from pyadvtools import standard_path
+
+from ..experiments_base import generate_standard_publisher_abbr_options_dict
+from ..spider.process_spider_bib import ProcessSpiderBib
+from ..spider.process_spider_url import ProcessSpiderUrl
+
+
+class CheckDeleteFormatMoveSpideredBibs(object):
+    """Experiment for check.
+
+    Args:
+        path_storage (str): path to storage
+        path_shutil (str): path to shutil
+        options (Dict[str, Any]): options for check, delete, format
+
+    Attributes:
+        path_storage (str): path to storage
+        path_shutil (str): path to shutil
+        publisher_abbr_dict (Dict[str, Dict[str, Any]]): publisher abbreviation options dict
+    """
+
+    def __init__(self, path_storage: str, path_shutil: str, options: Dict[str, Any]) -> None:
+        self.path_storage = standard_path(path_storage)
+        self.path_shutil = standard_path(path_shutil)
+        self.publisher_abbr_dict = generate_standard_publisher_abbr_options_dict(self.path_storage, options)
+
+    def check_delete_format_move(self) -> None:
+        publisher_abbr_dict = self.publisher_abbr_dict
+        for publisher in self.publisher_abbr_dict:
+            for abbr_standard in publisher_abbr_dict[publisher]:
+                self._check_format_check_move(publisher, abbr_standard, publisher_abbr_dict[publisher][abbr_standard])
+
+    def _check_format_check_move(self, publisher: str, abbr_standard: str, options: Dict[str, Any]) -> None:
+        path_abbr = os.path.join(self.path_storage, f"{publisher.lower()}/{abbr_standard}")
+
+        # for urls
+        if options.get("check_duplicate_url", False) is True:
+            ddu = options.get("delete_duplicate_url", False)
+            icdu = options.get("iterate_check_duplicate_url", False)
+            for extension in [".txt", ".csv"]:
+                ProcessSpiderUrl(path_abbr, abbr_standard).check_spider_url("url", extension, ddu, icdu)
+
+        # for bibs
+        if options.get("format_bib", False) is True:
+            wb = options.get("write_bib", False)
+            ProcessSpiderBib(path_abbr, abbr_standard).format_spider_bib(wb)
+
+        if options.get("check_duplicate_bib", False) is True:
+            ddb = options.get("delete_duplicate_bib", False)
+            ProcessSpiderBib(path_abbr, abbr_standard).check_spider_bib(ddb)
+
+        if options.get("move_bib", False) is True:
+            ps = os.path.join(self.path_shutil, publisher.lower())
+            ProcessSpiderBib(path_abbr, abbr_standard).move_spider_bib(ps)
+
+        # for early access
+        if options.get("early_access", False):
+            ProcessSpiderBib(path_abbr, abbr_standard).simplify_early_access()
+        return None

pyeasyphd/utils/utils.py

@@ -0,0 +1,62 @@
+import os
+import re
+
+from pyadvtools import (
+    combine_content_in_list,
+    read_list,
+    write_list,
+)
+
+html_head = """<!DOCTYPE html>
+<html>
+<head>
+  <meta charset="utf-8">
+  <title>{}</title>
+"""
+
+html_style = """  <style>
+    html {font-size: 22px;}
+    body {margin: 0 auto; max-width: 76em;}
+    #copyID {font-size: 18px;}
+  </style>
+  <script>
+    function copy(element) {
+      if (element.type == "button"){
+      element.type="text";
+      }
+      element.style.color="black";
+      element.style.backgroundColor="#C7EDCC";
+      element.select();
+      element.setSelectionRange(0, 99999);
+      navigator.clipboard.writeText(element.value);
+      window.getSelection().removeAllRanges();
+      element.type="button";
+    }
+  </script>
+</head>
+<body>
+"""
+
+html_tail = """
+</body>
+</html>
+"""
+
+textarea_header = '<textarea id="copyID" onclick="copy(this)" rows="16" cols="145">\n'
+textarea_tail = "\n</textarea>"
+
+
+def operate_on_generate_html(html_name: str) -> None:
+    if not (data_list := read_list(html_name, "r", None)):
+        return None
+
+    head_list = [html_head.format(os.path.basename(html_name).split('.')[0].strip()), html_style, "\n"]
+    tail_list = [html_tail]
+
+    content = "".join(data_list)
+    content = content.replace("<pre><code>", textarea_header).replace("</code></pre>", textarea_tail)
+    for i in re.findall(r"<li>(.*?)<details>", content, re.DOTALL):
+        content = content.replace(rf"<li>{i}<details>", f"<li><details>\n<summary>\n{i.strip()}\n</summary>")
+    data_list = combine_content_in_list([head_list, [content], tail_list])
+    write_list(data_list, html_name, "w", None, False)
+    return None

pyeasyphd-0.0.2.dist-info/METADATA

@@ -0,0 +1,27 @@
+Metadata-Version: 2.3
+Name: pyeasyphd
+Version: 0.0.2
+Summary: pyeasyphd
+License: GPL-3.0-or-later
+Keywords: Python,Markdown,LaTex
+Author: NextAI
+Author-email: nextartifintell@gmail.com
+Maintainer: NextAI
+Maintainer-email: nextartifintell@gmail.com
+Requires-Python: >=3.8.1
+Classifier: License :: OSI Approved :: GNU General Public License v3 or later (GPLv3+)
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Requires-Dist: pyadvtools (>=0.0.3,<0.0.4)
+Project-URL: Documentation, https://github.com/NextArtifIntell/pyeasyphd
+Project-URL: Homepage, https://github.com/NextArtifIntell/pyeasyphd
+Project-URL: Repository, https://github.com/NextArtifIntell/pyeasyphd
+Description-Content-Type: text/markdown
+
+# pyeasyphd
+

pyeasyphd-0.0.2.dist-info/RECORD

@@ -0,0 +1,80 @@
+pyeasyphd/.python-version,sha256=D6BK33IU5_SPJZB59_rpMy1cN6QwcHafPQUvGpjilS8,3
+pyeasyphd/Main.sublime-menu,sha256=6IKSvXyUSqq-eXrzSVGbzakbrPzw9n7hKe-p-FXmDXM,1561
+pyeasyphd/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+pyeasyphd/bib/__init__.py,sha256=lmagLbnGjT6-G_l-hUYxqjVJzgy-M_adA5tWcRRDOQc,22
+pyeasyphd/bib/bibtexbase/__init__.py,sha256=vUlnW7ZdoKIaWJpJ5wQWWln33YlQ9rqutwBq2FVOOcw,103
+pyeasyphd/bib/bibtexbase/standardize/_base.py,sha256=LEoU3XlaXAme6QVQtrb7_Q1_rmC9btr_qyLknF3T7gQ,1179
+pyeasyphd/bib/bibtexbase/standardize/default_data.py,sha256=g1O6PBv0XQXMiZiOadXbGGrWF42pw-2XM3YztkkwUnE,1661
+pyeasyphd/bib/bibtexbase/standardize/do_on_bib.py,sha256=1XBUBrregg_7l2XhTZkVfaBP3e5ApKEeGfHivg75nw0,1745
+pyeasyphd/bib/bibtexbase/standardize/do_on_comment_block.py,sha256=TTlkgtCxBUw7qfMdfB62R-zzgnQ0ypvRG2EpV3hFygM,1266
+pyeasyphd/bib/bibtexbase/standardize/do_on_entry_block.py,sha256=OfPMwWsmFLETY5uKl1QS7wgwFdngVabJlt5ywY3vRDk,11488
+pyeasyphd/bib/bibtexbase/standardize/do_on_preamble_block.py,sha256=D1nEEbzn_jbFLD3v1qgi8Kty13p-VF5jthBjNYHw3LQ,1194
+pyeasyphd/bib/bibtexbase/standardize/do_on_string_block.py,sha256=4ZUx6UuYc3J6uFuWGsR22Xl7UcmRqCIQc3S-tMuuXBY,1159
+pyeasyphd/bib/bibtexbase/standardize_bib.py,sha256=4uVL0z89sUzroAR1cyFy4nx4ARiVU3n52HCxOYJyXDQ,3020
+pyeasyphd/bib/bibtexparser/__init__.py,sha256=AuUumw0aSvRS9oUcTHd-RNKBvAAJHD-YdKEfAuKVubY,1215
+pyeasyphd/bib/bibtexparser/bibtex_format.py,sha256=uetBL8GFwrEqLJDe1_78MoY5lghfT9dZ6FwlgqKbv2E,3120
+pyeasyphd/bib/bibtexparser/exceptions.py,sha256=46MT8CHDj8pLq_951MWx-DXHWGo3zo9uZ58Zie7AjHc,2069
+pyeasyphd/bib/bibtexparser/library.py,sha256=Cw7nwxDr3k7sOAkhkbyHvtT6TY_xrSFN8n0TCJ57nbM,8341
+pyeasyphd/bib/bibtexparser/middlewares/block/add.py,sha256=UjabXwPYd6m9XDo2N3uHovQdTbmzbv5lCzW6XRcn-CQ,3300
+pyeasyphd/bib/bibtexparser/middlewares/block/authors.py,sha256=oggBBDg8ps5i6jLWad-JOKxKfa1cikTZUCFzwlx-qx0,863
+pyeasyphd/bib/bibtexparser/middlewares/block/doi_url.py,sha256=lfbO8_mtKsA4Jcl0FDcjQVOIuaWmjF-6BCDJ3I7WypY,2326
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_field_keys_normalize.py,sha256=DfT2EkwFSQ2svhz5rktKuZp2Q1seqgINYrBjKvQNppE,1915
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_field_keys_replace.py,sha256=2l4Tb5oAuVQjIU34CIwoKaLmhlPgFc9pMz6D02XpxWA,1020
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_field_values_normalize.py,sha256=r85H_eWm2HTuz2GnIJsM8NsEKGoUd692IKcWSh23fX8,9538
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_fields_delete.py,sha256=QWiyh1cBMxJmkGnDVcnxbxqSY17-ZA8P9LD2w4NjUYA,1069
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_fields_keep.py,sha256=Hqkne-SQX2VYsgpM0YGzfPs0X83LqDDbn2x6B7sjiuw,1277
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_fields_sort.py,sha256=YJCGdDmiFODLZe8Pzol0LYtSEokOWsQeGvdzqp10F8c,2521
+pyeasyphd/bib/bibtexparser/middlewares/block/entry_types.py,sha256=FhO_3gOQhSh0BUowi0Xn5jQHxwoBvXApOqIR8zm2dAs,541
+pyeasyphd/bib/bibtexparser/middlewares/block/journal_booktitle.py,sha256=cpTy9Yag__0UnPGhBuctvwBw79MilboqSbGPtXJsE64,4441
+pyeasyphd/bib/bibtexparser/middlewares/block/month_year.py,sha256=NXZF6kxixYc7yNc0_MkBVyWe5PaHLf3UfD5lO2CAv1g,1430
+pyeasyphd/bib/bibtexparser/middlewares/block/number_volume.py,sha256=1KmlBhLppjqww8OVmiDGgf7kXMD2VcGoBBVKOL0Q054,811
+pyeasyphd/bib/bibtexparser/middlewares/block/pages.py,sha256=71pWY1nuhV_d7aWkVeYU6iIAKnUY_3Q9w-KTzwJz83Y,1180
+pyeasyphd/bib/bibtexparser/middlewares/block/title.py,sha256=m_La5n795PT8LU-1cKubn57Bd6U0iChwUHmgX9rEcrM,727
+pyeasyphd/bib/bibtexparser/middlewares/library/generating_entrykeys.py,sha256=KospQjj7aO8FbY0HSlg35BlR9zPCXt1Ba2nILjQIEKo,3567
+pyeasyphd/bib/bibtexparser/middlewares/library/keeping_blocks.py,sha256=DpQxFdGXkF27x8ojV3DN8qwsBU1J44VsDcwV1AVR5u8,897
+pyeasyphd/bib/bibtexparser/middlewares/library/sorting_blocks.py,sha256=S8bQLuKcB9t0ox1Kq3GBXnhjhntKBrvuA7vVDHbkvaY,4758
+pyeasyphd/bib/bibtexparser/middlewares/middleware.py,sha256=mV4rn7aDJxqbBiAbo34xT2CsnNGewldcjWdE7Edu0dk,8860
+pyeasyphd/bib/bibtexparser/middlewares/parsestack.py,sha256=p1IG2glZyg0dywCYHhRx3eMC5GaX9IlepV3uzbFEb8s,464
+pyeasyphd/bib/bibtexparser/middlewares/utils.py,sha256=mLInKc651KIvkgt4_k73E6dz6WCmDpQmabxzwDKm3hk,4753
+pyeasyphd/bib/bibtexparser/middlewares_library_to_library.py,sha256=HopklPZl4Vm2xgRYFOA8hVSyW0BeeRwaNcCGVDIzCCk,20411
+pyeasyphd/bib/bibtexparser/middlewares_library_to_str.py,sha256=LZk06IyZFaga1WETU6IhqO3Lm7m-F5caWzGit3b-pOM,2008
+pyeasyphd/bib/bibtexparser/middlewares_str_to_library.py,sha256=931mKNxiP6zlBFtYP8xHd7x2n12My01dF_VWGIXySIU,1090
+pyeasyphd/bib/bibtexparser/middlewares_str_to_str.py,sha256=2JcfiYDk6dpJ51-EUwKUjG6-WAyI3qDgCalpbxbdGVg,1086
+pyeasyphd/bib/bibtexparser/model.py,sha256=YSMUznMRRsNGThRKjof6JeEyFrSpEG9dw9FUM6x-drw,15346
+pyeasyphd/bib/bibtexparser/splitter.py,sha256=waiNi5jnN6q8Iwbtzgy3soYKCEMU_1433VLc62TMM9k,6365
+pyeasyphd/bib/core/__init__.py,sha256=YRclQ4sZJFwOujh2eyhfhytui0LiTGNX6xzz90BuLiQ,662
+pyeasyphd/bib/core/convert_library_to_library.py,sha256=kSiFVkGyANUGX4e1zAy-L4YyPo4-e1rjnCWkN1a_dm4,1092
+pyeasyphd/bib/core/convert_library_to_str.py,sha256=yvea1NUobcn2NF9CtvrjmNN1IU23z7bJZ_G3-QwyWQI,7303
+pyeasyphd/bib/core/convert_str_to_library.py,sha256=CNB5UfSOcCqPPypxm9yIY7R7kvLQ_fzRPIqKwdegToY,1017
+pyeasyphd/bib/core/convert_str_to_str.py,sha256=jUxXD3v7t042OaMXPFlWGHjNNukpuoel12TpYt1krD4,900
+pyeasyphd/main/__init__.py,sha256=7_x6_RNSeha2zrgIbKjJgEvFWK62btOW9v6RUU1Ygm8,390
+pyeasyphd/main/basic_input.py,sha256=Oh-JG4BtdVimJHtINq4zkfvN4av1bJ39ZErmJISaJHE,6767
+pyeasyphd/main/pandoc_md_to.py,sha256=hRajMqlPl879qQKOE-MLtCE-KL9U5y2GO6et8cgkm8w,14806
+pyeasyphd/main/python_run_bib.py,sha256=RRVeGXXs1MELpAYex9vT3x97Rqr211TP1aeHhD-8hXU,2891
+pyeasyphd/main/python_run_md.py,sha256=6K4zqYtl298-iS1XVGH4jY9FkDD9ezwpGfZk0NIgk2Q,10353
+pyeasyphd/main/python_run_tex.py,sha256=LQirSh_5Mv5t-SdAtV5Ge2aCTTkvaPNAloFTH6hkzCI,5691
+pyeasyphd/main/python_writers.py,sha256=VQpaAbdNNa1s327ADgl8Wy-rKBZy1alqmtmx6E3RDWI,8510
+pyeasyphd/pyeasyphd.py,sha256=vPM4uPmr6jSxVOy0zYetoijc8wUkQdXExUVdJyWynVQ,2875
+pyeasyphd/pyeasyphd.sublime-settings,sha256=ZtclkJnOoZUmCLhWWsEDu3eiHyHv1JGP7g6gOMG6yEY,8664
+pyeasyphd/pyeasyphd.sublime-syntax,sha256=pXylbA-tye-K5dCTjEJLFVRqtY1T7AgWZ4laxo-dnaE,73
+pyeasyphd/tools/__init__.py,sha256=4l2e19MWmVrAvIH3IOA7v1EPaaBBvuq1UbXhmd6_bts,1041
+pyeasyphd/tools/compare/compare_bibs.py,sha256=QZlqVHbF133nnZZpzDa6jYlkHnai_x7kqgq62hpIzw0,10156
+pyeasyphd/tools/experiments_base.py,sha256=mx-vt0dpgjhsYcRyVpQkzdbJw_CFXcaaYt81w13y4Tc,7639
+pyeasyphd/tools/format_save_bibs.py,sha256=vChgzsg45MT6IkeN8ZuJhMdGDMIvwmH8NGr16IuQwy4,7043
+pyeasyphd/tools/generate/generate_from_bibs.py,sha256=RP-ADsmUsM6lEUaHDMAv6DSVrsx5wHyHsRVDVZwjaeM,17794
+pyeasyphd/tools/generate/generate_links.py,sha256=lnPf8FmksPDBtb30wHarr7y9cdFm3ELnKZwXIKHukyw,13700
+pyeasyphd/tools/py_run_bib_md_tex.py,sha256=lc6JqfXeJSMA8icTJTC7X1pq4tgurOaXq9f0-LSJK7g,14657
+pyeasyphd/tools/replace/replace.py,sha256=wXdO72PpqZ-Iyzq5AjVCIpx8h2d1YCTiWY9VSbQAVvs,3337
+pyeasyphd/tools/search/data.py,sha256=7uHXfHTcn88zkfqAGZMNcSd2OjS2LhkpE9paNuSUUqw,10754
+pyeasyphd/tools/search/search_base.py,sha256=Q8IWqVLk-rGZwRZ316WkZNs0qyF7PVHP6lx6Ul5mLKI,4507
+pyeasyphd/tools/search/search_core.py,sha256=16nVB2j3mtrTxUJKgtYg65ux6bI81ofTHn1IAB3twPc,13783
+pyeasyphd/tools/search/search_keywords.py,sha256=mNlwLWSgi19KiErCb5IB433L96sBWVFWgjd5IO_GQuU,10372
+pyeasyphd/tools/search/search_writers.py,sha256=TuQGm9QwHy0bZHNtO_qg-MYEcip_rUPfwHyvPKO-sEg,12551
+pyeasyphd/tools/search/utils.py,sha256=yLKrMmM2x-fHdm6P8haFZzagd9sLkViTdwJTHyGRgXY,6092
+pyeasyphd/tools/spider/process_spider_bib.py,sha256=p1UNUaLAxMZPYtuwjNi20e1ms--cyjlB7CIZFP3SX6U,10443
+pyeasyphd/tools/spider/process_spider_url.py,sha256=YKG4GER5TiFDjyk94n1ocUQ8bM7ZvXImx0lKNvnyVC4,2970
+pyeasyphd/tools/spider/process_spider_url_bib.py,sha256=1b0FnUlTt5BzaOqFZJmnY5gnAhdD4pbQolHR1e3wiyE,2705
+pyeasyphd/utils/utils.py,sha256=yuHy6D2mTsd2aAwNBSMC7CU4LV-MkWwzKeO_0Iehbyg,1655
+pyeasyphd-0.0.2.dist-info/METADATA,sha256=pK0O12eHXWGFLRBzOsmyxyd4hXmjVjGBT9j-I1D-blg,1043
+pyeasyphd-0.0.2.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
+pyeasyphd-0.0.2.dist-info/RECORD,,

pyeasyphd-0.0.2.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: poetry-core 2.1.3
+Root-Is-Purelib: true
+Tag: py3-none-any