routers/tool_find_related.py

# routers/find_related.py

import os
import pickle
import torch
import re

from typing import List
from datetime import datetime, timedelta
from enum import Enum
from sentence_transformers import util
from fastapi import APIRouter
from fastapi.responses import PlainTextResponse

try:
    from .rag import EMBEDDING_CTX
    from .utils_gitea import gitea_fetch_issues, gitea_json_issue_get, gitea_issues_body_updated_at_get
except:
    from rag import EMBEDDING_CTX
    from utils_gitea import gitea_fetch_issues, gitea_json_issue_get, gitea_issues_body_updated_at_get


router = APIRouter()

issue_attr_filter = {'number', 'title', 'body',
                     'state', 'updated_at', 'created_at'}


class State(str, Enum):
    opened = "opened"
    closed = "closed"
    all = "all"


class _Data(dict):
    cache_path = "routers/rag/embeddings_issues.pkl"

    @staticmethod
    def _create_issue_string(title, body):
        cleaned_body = body.replace('\r', '')
        cleaned_body = cleaned_body.replace('**System Information**\n', '')
        cleaned_body = cleaned_body.replace('**Blender Version**\n', '')
        cleaned_body = cleaned_body.replace(
            'Worked: (newest version of Blender that worked as expected)\n', '')
        cleaned_body = cleaned_body.replace(
            '**Short description of error**\n', '')
        cleaned_body = cleaned_body.replace('**Addon Information**\n', '')
        cleaned_body = cleaned_body.replace(
            '**Exact steps for others to reproduce the error**\n', '')
        cleaned_body = cleaned_body.replace(
            '[Please describe the exact steps needed to reproduce the issue]\n', '')
        cleaned_body = cleaned_body.replace(
            '[Please fill out a short description of the error here]\n', '')
        cleaned_body = cleaned_body.replace(
            '[Based on the default startup or an attached .blend file (as simple as possible)]\n', '')
        cleaned_body = re.sub(
            r', branch: .+?, commit date: \d{4}-\d{2}-\d{2} \d{2}:\d{2}, hash: `.+?`', '', cleaned_body)
        cleaned_body = re.sub(
            r'\/?attachments\/[a-zA-Z0-9\-]+', 'attachment', cleaned_body)
        cleaned_body = re.sub(
            r'https?:\/\/[^\s/]+(?:\/[^\s/]+)*\/([^\s/]+)', lambda match: match.group(1), cleaned_body)

        return title + '\n' + cleaned_body

    @staticmethod
    def _find_latest_date(issues, default_str=None):
        # Handle the case where 'issues' is empty
        if not issues:
            return default_str

        return max((issue['updated_at'] for issue in issues), default=default_str)

    @classmethod
    def _create_strings_to_embbed(cls, issues):
        texts_to_embed = [cls._create_issue_string(
            issue['title'], issue['body']) for issue in issues]

        return texts_to_embed

    def _data_ensure_size(self, repo, size_new):
        ARRAY_CHUNK_SIZE = 4096

        updated_at_old = None
        arrays_size_old = 0
        titles_old = []
        try:
            arrays_size_old = self[repo]['arrays_size']
            titles_old = self[repo]['titles']
            if size_new <= arrays_size_old:
                return
        except:
            pass

        arrays_size_new = ARRAY_CHUNK_SIZE * \
            (int(size_new / ARRAY_CHUNK_SIZE) + 1)

        data_new = {
            'updated_at': updated_at_old,
            'arrays_size': arrays_size_new,
            'titles': titles_old + [None] * (arrays_size_new - arrays_size_old),
            'embeddings': torch.empty((arrays_size_new, *EMBEDDING_CTX.embedding_shape),
                                      dtype=EMBEDDING_CTX.embedding_dtype,
                                      device=EMBEDDING_CTX.embedding_device),
            'opened':  torch.zeros(arrays_size_new, dtype=torch.bool),
            'closed':  torch.zeros(arrays_size_new, dtype=torch.bool),
        }

        try:
            data_new['embeddings'][:arrays_size_old] = self[repo]['embeddings']
            data_new['opened'][:arrays_size_old] = self[repo]['opened']
            data_new['closed'][:arrays_size_old] = self[repo]['closed']
        except:
            pass

        self[repo] = data_new

    def _embeddings_generate(self, repo):
        if os.path.exists(self.cache_path):
            with open(self.cache_path, 'rb') as file:
                data = pickle.load(file)
                self.update(data)
                if repo in self:
                    return

        issues = gitea_fetch_issues('blender', repo, state='all', since=None,
                                    issue_attr_filter=issue_attr_filter)

        # issues = sorted(issues, key=lambda issue: int(issue['number']))

        print("Embedding Issues...")
        texts_to_embed = self._create_strings_to_embbed(issues)
        embeddings = EMBEDDING_CTX.encode(texts_to_embed)

        self._data_ensure_size(repo, int(issues[0]['number']))
        self[repo]['updated_at'] = self._find_latest_date(issues)

        titles = self[repo]['titles']
        embeddings_new = self[repo]['embeddings']
        opened = self[repo]['opened']
        closed = self[repo]['closed']

        for i, issue in enumerate(issues):
            number = int(issue['number'])
            titles[number] = issue['title']
            embeddings_new[number] = embeddings[i]
            if issue['state'] == 'open':
                opened[number] = True
            if issue['state'] == 'closed':
                closed[number] = True

    def _embeddings_updated_get(self, repo):
        with EMBEDDING_CTX.lock:
            if not repo in self:
                self._embeddings_generate(repo)

            date_old = self[repo]['updated_at']

            issues = gitea_fetch_issues(
                'blender', repo, since=date_old, issue_attr_filter=issue_attr_filter)

            # Get the most recent date
            date_new = self._find_latest_date(issues, date_old)

            if date_new == date_old:
                # Nothing changed
                return self[repo]

            self[repo]['updated_at'] = date_new

    # autopep8: off
            # Consider that if the time hasn't changed, it's the same issue.
            issues = [issue for issue in issues if issue['updated_at'] != date_old]

            self._data_ensure_size(repo, int(issues[0]['number']))

            updated_at = gitea_issues_body_updated_at_get(issues)
            issues_to_embed = []

            for i, issue in enumerate(issues):
                number = int(issue['number'])
                self[repo]['opened'][number] = issue['state'] == 'open'
                self[repo]['closed'][number] = issue['state'] == 'closed'

                title_old = self[repo]['titles'][number]
                if title_old != issue['title']:
                    self[repo]['titles'][number] = issue['title']
                    issues_to_embed.append(issue)
                elif not updated_at[i] or updated_at[i] >= date_old:
                    issues_to_embed.append(issue)

            if issues_to_embed:
                print(f"Embedding {len(issues_to_embed)} issue{'s' if len(issues_to_embed) > 1 else ''}")
                texts_to_embed = self._create_strings_to_embbed(issues_to_embed)
                embeddings = EMBEDDING_CTX.encode(texts_to_embed)

                for i, issue in enumerate(issues_to_embed):
                    number = int(issue['number'])
                    self[repo]['embeddings'][number] = embeddings[i]

    # autopep8: on
        return self[repo]

    def _sort_similarity(self,
                         repo: str,
                         query_emb: List[torch.Tensor],
                         limit: int,
                         state: State = State.opened) -> list:
        duplicates = []

        data = self[repo]
        embeddings = data['embeddings']
        mask_opened = data["opened"]

        if state == State.all:
            mask = mask_opened | data["closed"]
        else:
            mask = data[state.value]

        embeddings = embeddings[mask]
        true_indices = mask.nonzero(as_tuple=True)[0]

        ret = util.semantic_search(
            query_emb, embeddings, top_k=limit, score_function=util.dot_score)

        for score in ret[0]:
            corpus_id = score['corpus_id']
            number = true_indices[corpus_id].item()
            closed_char = "" if mask_opened[number] else "~~"
            text = f"{closed_char}#{number}{closed_char}: {data['titles'][number]}"
            duplicates.append(text)

        return duplicates

    def find_relatedness(self, repo: str, number: int, limit: int = 20, state: State = State.opened):
        data = self._embeddings_updated_get(repo)

        # Check if the embedding already exists.
        if data['titles'][number] is not None:
            new_embedding = data['embeddings'][number]
        else:
            gitea_issue = gitea_json_issue_get('blender', repo, number)
            text_to_embed = self._create_issue_string(
                gitea_issue['title'], gitea_issue['body'])

            new_embedding = EMBEDDING_CTX.encode([text_to_embed])

        duplicates = self._sort_similarity(
            repo, new_embedding, limit=limit, state=state)

        if not duplicates:
            return ''

        if match := re.search(r'(~~)?#(\d+)(~~)?:', duplicates[0]):
            number_cached = int(match.group(2))
            if number_cached == number:
                return '\n'.join(duplicates[1:])

        return '\n'.join(duplicates)


G_data = _Data()


@router.get("/find_related/{repo}/{number}", response_class=PlainTextResponse)
def find_related(repo: str = 'blender', number: int = 104399, limit: int = 15, state: State = State.opened) -> str:
    related = G_data.find_relatedness(repo, number, limit=limit, state=state)
    return related


if __name__ == "__main__":
    update_cache = True
    if update_cache:
        G_data._embeddings_updated_get('blender')
        G_data._embeddings_updated_get('blender-addons')
        with open(G_data.cache_path, "wb") as file:
            # Converting the embeddings to be CPU compatible, as the virtual machine in use currently only supports the CPU.
            for val in G_data.values():
                val['embeddings'] = val['embeddings'].to(torch.device('cpu'))

            pickle.dump(dict(G_data), file, protocol=pickle.HIGHEST_PROTOCOL)

    # Converting the embeddings to be GPU.
    for val in G_data.values():
        val['embeddings'] = val['embeddings'].to(torch.device('cuda'))

    # 'blender/blender/111434' must print #96153, #83604 and #79762
    related1 = G_data.find_relatedness(
        'blender', 111434, limit=20, state=State.all)
    related2 = G_data.find_relatedness('blender-addons', 104399, limit=20)

    print("These are the 20 most related issues:")
    print(related1)
    print()
    print("These are the 20 most related issues:")
    print(related2)