logdetective 2.0.1__py3-none-any.whl → 2.11.0__py3-none-any.whl

logdetective/server/templates/base_response.html.j2

@@ -0,0 +1,59 @@
+<b>Log Detective report
+    {% if project %}for {{ project }}{% endif %}
+</b>
+<p>
+    Analysis:
+    <br>
+    {{ explanation }}
+    <br>
+    <details>
+        <summary>Analysis details</summary>
+        <ul>
+            {% for snippet in snippets %}
+                <li>
+                    <b>Line {{ snippet.line_number }}:</b> <code>{{ snippet.text }}</code>
+                    {{ snippet.explanation.text }}
+                </li>
+            {% endfor %}
+        </ul>
+    </details>
+</p>
+<br>
+<details>
+    <summary>Help</summary>
+    <p>
+        Log Detective analyses logs using template mining algorithm and LLMs.
+        Providing overview of the primary issue, while highlighting potentially important log lines.
+    </p>
+    <p>
+        Log Detective may offer potential solution to problem.
+        However, it is designed to do so only when sufficiently confident in the response.
+        <br>
+        Even so, as with any LLM based AI system, recommendations of Log Detective must be taken with caution.
+        Before acting on any recommendations of Log Detective, consider following:
+    </p>
+    <ul>
+        <li>
+            <b>Do you understand how does the proposed solution work? And why should it solve your problem?</b>
+        </li>
+        <li>Did Log Detective work with sufficient information to make this suggestion?</li>
+        <li>Are the suggested actions safe in the context of your work?</li>
+        <li>Do results of snippet analysis align with results of final analysis?</li>
+        <li>Are there any potentially adverse outcomes from implementing proposed solution?</li>
+    </ul>
+</details>
+<br>
+{% render_contacts %}
+<details>
+    <summary>Contacts</summary>
+    <p>Don't hesitate to reach out with your questions or suggestions.</p>
+    <ul>
+        <li>
+            <a href="https://github.com/fedora-copr/logdetective">Source</a>
+        </li>
+        <li>
+            <a href="https://github.com/fedora-copr/logdetective/issues">Issue tracker</a>
+        </li>
+    </ul>
+</details>
+{% endif %}

logdetective/server/templates/gitlab_full_comment.md.j2

@@ -1,68 +1,73 @@
-The package {{ package }} failed to build, here is a possible explanation why.
-
+The package {{ package }} build has experienced an issue.
 Please know that the explanation was provided by AI and may be incorrect.
 {% if certainty > 0 %}
-In this case, we are {{ "%.2f" | format(certainty) }}% certain of the response {{ emoji_face }}.
+    In this case, we are {{ "%.2f" | format(certainty) }}% certain of the response {{ emoji_face }}.
 {% endif %}
 
-{{ explanation }}
-
+<details open>
+    <summary>Description</summary>
+    {{ explanation }}
+</details>
 <details>
+{#
+    Formatted so that we don't trigger GitLab markdown
+#}
+<summary>Snippets</summary>
 <ul>
-{% for snippet in snippets %}
-<li>
-<b>Line {{ snippet.line_number }}:</b> <code>{{ snippet.text }}</code>
-{{ snippet.explanation.text }}
-</li>
-{% endfor %}
+{% for snippet in snippets -%}
+<li><div><b>Line {{ snippet.line_number }}:</b> <code>{{ snippet.text | e }}</code><br>{{ snippet.explanation.text | e }}</div></li>
+{%- endfor %}
 </ul>
 </details>
-
 <details>
-  <summary>Logs</summary>
-  <p>
-    Log Detective analyzed the following logs files to provide an explanation:
-  </p>
-
-  <ul>
-    <li><a href="{{ log_url }}">{{ log_url }}</a></li>
-  </ul>
-
-  <p>
-    Additional logs are available from:
+    <summary>Logs</summary>
+    <p>Log Detective analyzed the following logs files to provide an explanation:</p>
     <ul>
-    <li><a href="{{ artifacts_url }}">artifacts.zip</a></li>
-  </ul>
-  </p>
-
-  <p>
-    Please know that these log files are automatically removed after some
-    time, so you might need a backup.
-  </p>
+        <li>
+            <a href="{{ log_url }}">{{ log_url }}</a>
+        </li>
+    </ul>
+    <p>
+        Additional logs are available from:
+        <ul>
+            <li>
+                <a href="{{ artifacts_url }}">artifacts.zip</a>
+            </li>
+        </ul>
+    </p>
+    <p>
+        Please know that these log files are automatically removed after some
+        time, so you might need a backup.
+    </p>
 </details>
-
 <details>
-  <summary>Help</summary>
-  <p>Don't hesitate to reach out.</p>
-
-  <ul>
-    <li><a href="https://github.com/fedora-copr/logdetective">Upstream</a></li>
-    <li><a href="https://github.com/fedora-copr/logdetective/issues">Issue tracker</a></li>
-    <li><a href="https://redhat.enterprise.slack.com/archives/C06DWNVKKDE">Slack</a></li>
-    <li><a href="https://log-detective.com/documentation">Documentation</a></li>
-  </ul>
+    <summary>Help</summary>
+    <p>Don't hesitate to reach out.</p>
+    <ul>
+        <li>
+            <a href="https://github.com/fedora-copr/logdetective">Upstream</a>
+        </li>
+        <li>
+            <a href="https://github.com/fedora-copr/logdetective/issues">Issue tracker</a>
+        </li>
+        <li>
+            <a href="https://redhat.enterprise.slack.com/archives/C06DWNVKKDE">Slack</a>
+        </li>
+        <li>
+            <a href="https://log-detective.com/documentation">Documentation</a>
+        </li>
+    </ul>
 </details>
 
+<hr>
 
----
-This comment was created by [Log Detective][log-detective].
-
-Was the provided feedback accurate and helpful? <br>Please vote with :thumbsup:
-or :thumbsdown: to help us improve.<br>
-
-<i>If this Log Detective report contains harmful content, please use the
-[Gitlab reporting feature for harmful content](https://docs.gitlab.com/user/report_abuse/)
-and contact the [Log Detective developers](https://github.com/fedora-copr/logdetective/issues).</i>
-
-[log-detective]: https://log-detective.com/
-[contact]: https://github.com/fedora-copr
+This explanation was provided by AI (<a href="https://logdetective.com">Log Detective</a>).
+Always review AI generated content prior to use.
+Was the provided feedback accurate and helpful?
+<br>
+Please vote with :thumbsup:
+or :thumbsdown: to help us improve.
+<br>
+<i>If this Log Detective report contains harmful content,
+please use the <a href="https://docs.gitlab.com/user/report_abuse/">Gitlab reporting feature for harmful content</a>
+and contact the <a href="https://github.com/fedora-copr/logdetective/issues">Log Detective developers</a>.</i>

logdetective/server/templates/gitlab_short_comment.md.j2

@@ -1,57 +1,62 @@
-The package {{ package }} failed to build, here is a possible explanation why.
-
+The package {{ package }} build has experienced an issue.
 Please know that the explanation was provided by AI and may be incorrect.
 {% if certainty > 0 %}
-In this case, we are {{ "%.2f" | format(certainty) }}% certain of the response {{ emoji_face }}.
+    In this case, we are {{ "%.2f" | format(certainty) }}% certain of the response {{ emoji_face }}.
 {% endif %}
 
-{{ explanation }}
-
+<details open>
+    <summary>Description</summary>
+    {{ explanation }}
+</details>
 <details>
-  <summary>Logs</summary>
-  <p>
-    Log Detective analyzed the following logs files to provide an explanation:
-  </p>
-
-  <ul>
-    <li><a href="{{ log_url }}">{{ log_url }}</a></li>
-  </ul>
-
-  <p>
-    Additional logs are available from:
+    <summary>Logs</summary>
+    <p>Log Detective analyzed the following logs files to provide an explanation:</p>
     <ul>
-    <li><a href="{{ artifacts_url }}">artifacts.zip</a></li>
-  </ul>
-  </p>
-
-  <p>
-    Please know that these log files are automatically removed after some
-    time, so you might need a backup.
-  </p>
+        <li>
+            <a href="{{ log_url }}">{{ log_url }}</a>
+        </li>
+    </ul>
+    <p>
+        Additional logs are available from:
+        <ul>
+            <li>
+                <a href="{{ artifacts_url }}">artifacts.zip</a>
+            </li>
+        </ul>
+    </p>
+    <p>
+        Please know that these log files are automatically removed after some
+        time, so you might need a backup.
+    </p>
 </details>
-
 <details>
-  <summary>Help</summary>
-  <p>Don't hesitate to reach out.</p>
-
-  <ul>
-    <li><a href="https://github.com/fedora-copr/logdetective">Upstream</a></li>
-    <li><a href="https://github.com/fedora-copr/logdetective/issues">Issue tracker</a></li>
-    <li><a href="https://redhat.enterprise.slack.com/archives/C06DWNVKKDE">Slack</a></li>
-    <li><a href="https://log-detective.com/documentation">Documentation</a></li>
-  </ul>
+    <summary>Help</summary>
+    <p>Don't hesitate to reach out.</p>
+    <ul>
+        <li>
+            <a href="https://github.com/fedora-copr/logdetective">Upstream</a>
+        </li>
+        <li>
+            <a href="https://github.com/fedora-copr/logdetective/issues">Issue tracker</a>
+        </li>
+        <li>
+            <a href="https://redhat.enterprise.slack.com/archives/C06DWNVKKDE">Slack</a>
+        </li>
+        <li>
+            <a href="https://log-detective.com/documentation">Documentation</a>
+        </li>
+    </ul>
 </details>
 
-
----
-This comment was created by [Log Detective][log-detective].
-
-Was the provided feedback accurate and helpful? <br>Please vote with :thumbsup:
-or :thumbsdown: to help us improve.<br>
-
-<i>If this Log Detective report contains harmful content, please use the
-[Gitlab reporting feature for harmful content](https://docs.gitlab.com/user/report_abuse/)
-and contact the [Log Detective developers](https://github.com/fedora-copr/logdetective/issues).</i>
-
-[log-detective]: https://log-detective.com/
-[contact]: https://github.com/fedora-copr
+<hr>
+
+This explanation was provided by AI (<a href="https://logdetective.com">Log Detective</a>).
+Always review AI generated content prior to use.
+Was the provided feedback accurate and helpful?
+<br>
+Please vote with :thumbsup:
+or :thumbsdown: to help us improve.
+<br>
+<i>If this Log Detective report contains harmful content,
+please use the <a href="https://docs.gitlab.com/user/report_abuse/">Gitlab reporting feature for harmful content</a>
+and contact the <a href="https://github.com/fedora-copr/logdetective/issues">Log Detective developers</a>.</i>

logdetective/server/utils.py

@@ -1,15 +1,11 @@
-from typing import List, Tuple
+from typing import List
+from importlib.metadata import version
 
 import aiohttp
 from fastapi import HTTPException
 
 from logdetective.constants import SNIPPET_DELIMITER
-from logdetective.extractors import DrainExtractor
-from logdetective.server.config import (
-    LOG,
-    SERVER_CONFIG,
-    SKIP_SNIPPETS_CONFIG,
-)
+from logdetective.server.config import LOG
 from logdetective.server.exceptions import LogDetectiveConnectionError
 from logdetective.server.models import AnalyzedSnippet, RatedSnippetAnalysis
 
@@ -22,26 +18,6 @@ def format_analyzed_snippets(snippets: list[AnalyzedSnippet]) -> str:
     return summary
 
 
-def mine_logs(log: str) -> List[Tuple[int, str]]:
-    """Extract snippets from log text"""
-    extractor = DrainExtractor(
-        verbose=True,
-        context=True,
-        max_clusters=SERVER_CONFIG.extractor.max_clusters,
-        skip_snippets=SKIP_SNIPPETS_CONFIG,
-        max_snippet_len=SERVER_CONFIG.extractor.max_snippet_len
-    )
-
-    LOG.info("Getting summary")
-    log_summary = extractor(log)
-
-    ratio = len(log_summary) / len(log.split("\n"))
-    LOG.debug("Log summary: \n %s", log_summary)
-    LOG.info("Compression ratio: %s", ratio)
-
-    return log_summary
-
-
 def connection_error_giveup(details: dict) -> None:
     """Too many connection errors, give up.
     """
@@ -120,3 +96,15 @@ def filter_snippets(
     processed_snippets = sorted(processed_snippets, key=select_line_number)
 
     return processed_snippets
+
+
+def construct_final_prompt(formatted_snippets: str, prompt_template: str) -> str:
+    """Create final prompt from processed snippets and csgrep output, if it is available."""
+
+    final_prompt = prompt_template.format(formatted_snippets)
+    return final_prompt
+
+
+def get_version() -> str:
+    """Obtain the version number using importlib"""
+    return version('logdetective')

logdetective/utils.py

@@ -1,5 +1,6 @@
 import logging
 import os
+import subprocess as sp
 from typing import Iterator, List, Dict, Tuple, Generator
 from urllib.parse import urlparse
 
@@ -7,60 +8,73 @@ import aiohttp
 import numpy as np
 import yaml
 
-from llama_cpp import Llama, CreateCompletionResponse, CreateCompletionStreamResponse
+from llama_cpp import (
+    Llama,
+    CreateChatCompletionResponse,
+    CreateChatCompletionStreamResponse,
+)
+from logdetective.constants import SNIPPET_DELIMITER
 from logdetective.models import PromptConfig, SkipSnippets
 from logdetective.remote_log import RemoteLog
 
-
 LOG = logging.getLogger("logdetective")
 
 
-def chunk_continues(text: str, index: int) -> bool:
+def new_message(text: str) -> bool:
     """Set of heuristics for determining whether or not
     does the current chunk of log text continue on next line.
 
     Following rules are checked, in order:
-    * is the next character is whitespace
-    * is the previous character backslash '\\'
-    * is the previous character colon ':'
-
+    * is the first character is whitespace
+    * is the first character backslash '|'
     """
     conditionals = [
-        lambda i, string: string[i + 1].isspace(),
-        lambda i, string: string[i - 1] == "\\",
-        lambda i, string: string[i - 1] == ":",
+        lambda string: string[0].isspace(),
+        lambda string: string[0] == "|",
     ]
 
     for c in conditionals:
-        y = c(index, text)
+        y = c(text)
         if y:
-            return True
+            return False
 
-    return False
+    return True
 
 
-def get_chunks(text: str, max_len: int = 2000) -> Generator[Tuple[int, str], None, None]:
+def get_chunks(
+    text: str, max_chunk_len: int = 2000
+) -> Generator[Tuple[int, str], None, None]:
     """Split log into chunks according to heuristic
     based on whitespace and backslash presence.
     """
-    text_len = len(text)
-    i = 0
+    lines = text.splitlines()
+
+    # Chunk we will be yielding
     chunk = ""
-    # Keep track of the original and next line number
-    # every `\n` hit increases the next_line_number by one.
-    original_line_number = 0
-    next_line_number = 0
-    while i < text_len:
-        chunk += text[i]
-        if text[i] == "\n":
-            next_line_number += 1
-            if i + 1 < text_len and chunk_continues(text, i) and i + 1 < max_len:
-                i += 1
-                continue
-            yield (original_line_number, chunk)
-            original_line_number = next_line_number + 1
-            chunk = ""
-        i += 1
+    # Number of line where the message started
+    original_line = 0
+    for i, line in enumerate(lines):
+        if len(line) == 0:
+            continue
+        if new_message(line):
+            # Yield chunk if we have it
+            if len(chunk) > 0:
+                yield (original_line, chunk)
+            original_line = i
+            chunk = line
+        else:
+            chunk += "\n" + line
+        if len(chunk) > max_chunk_len:
+            # If the chunk is too long, keep splitting into smaller chunks
+            # until we reach manageable size
+            while len(chunk) > max_chunk_len:
+                remainder = chunk[max_chunk_len:]
+                chunk = chunk[:max_chunk_len]
+                yield (original_line, chunk)
+                chunk = remainder
+
+    # if we still have some text left over
+    yield (original_line, chunk)
 
 
 def initialize_model(
@@ -113,8 +127,8 @@ def compute_certainty(probs: List[Dict]) -> float:
 
 
 def process_log(
-    log: str, model: Llama, stream: bool, prompt_template: str, temperature: float
-) -> CreateCompletionResponse | Iterator[CreateCompletionStreamResponse]:
+    log: str, model: Llama, stream: bool, prompt_templates: PromptConfig, temperature: float
+) -> CreateChatCompletionResponse | Iterator[CreateChatCompletionStreamResponse]:
     """Processes a given log using the provided language model and returns its summary.
 
     Args:
@@ -126,11 +140,23 @@ def process_log(
     Returns:
         str: The summary of the given log generated by the language model.
     """
-    response = model(
-        prompt=prompt_template.format(log),
+    messages = [
+        {
+            "role": "system",
+            "content": prompt_templates.default_system_prompt
+        },
+        {
+            "role": "user",
+            "content": prompt_templates.prompt_template.format(log)
+        },
+    ]
+
+    response = model.create_chat_completion(
+        messages=messages,
         stream=stream,
         max_tokens=0,
-        logprobs=1,
+        logprobs=True,
+        top_logprobs=1,
         temperature=temperature,
     )
 
@@ -166,22 +192,21 @@ def format_snippets(snippets: list[str] | list[Tuple[int, str]]) -> str:
     Line number must be first element in the tuple. Mixed format of snippets
     is permitted, but may have impact on inference.
     """
-    summary = ""
+    summary = "\n"
     for i, s in enumerate(snippets):
         if isinstance(s, tuple):
-            summary += f"""
-            Snippet No. {i} at line #{s[0]}:
-
-            {s[1]}
-            ================
-            """
+            line_number, snippet_content = s
+            header = f"Snippet No. {i} at line #{line_number}:"
         else:
-            summary += f"""
-            Snippet No. {i}:
-
-            {s}
-            ================
-            """
+            header = f"Snippet No. {i}:"
+            snippet_content = s
+        summary += (
+            f"{header}\n"
+            "\n"
+            f"{snippet_content}\n"
+            f"{SNIPPET_DELIMITER}\n"
+            f"\n"
+        )
     return summary
 
 
@@ -247,3 +272,44 @@ def load_skip_snippet_patterns(path: str | None) -> SkipSnippets:
             raise e
 
     return SkipSnippets({})
+
+
+def check_csgrep() -> bool:
+    """Verifies presence of csgrep in path"""
+    try:
+        result = sp.run(
+            ["csgrep", "--version"],
+            text=True,
+            check=True,
+            shell=False,
+            capture_output=True,
+            timeout=1.0,
+        )
+    except (FileNotFoundError, sp.TimeoutExpired, sp.CalledProcessError) as ex:
+        LOG.error("Required binary `csgrep` was not found in path: %s", ex)
+        return False
+    if result.returncode == 0:
+        return True
+    LOG.error("Issue was encountered while calling `csgrep`: `%s`", result.stderr)
+
+    return False
+
+
+def mine_logs(log: str, extractors: list) -> List[Tuple[int, str]]:
+    """Extract snippets from log text using extractors provided.
+    Each extractor is applied in turn on original log.
+    Depending on characteristics of extractors used, there may be
+    an overlap in snippets extracted."""
+
+    log_summary = []
+
+    LOG.info("Getting summary")
+
+    for extractor in extractors:
+        log_summary.extend(extractor(log))
+
+    ratio = len("\n".join([text for _, text in log_summary])) / len(log)
+    LOG.debug("Log summary: \n %s", log_summary)
+    LOG.info("Snippets: %s Compression ratio: %s", len(log_summary), ratio)
+
+    return log_summary