deepresearch-flow 0.1.1__py3-none-any.whl

deepresearch_flow/paper/templates/default_paper.md.j2

@@ -0,0 +1,32 @@
+# {{ paper_title }}
+
+{% set is_zh = output_language == "zh" %}
+**{{ "作者 / Authors" if is_zh else "Authors" }}:** {{ paper_authors | join(", ") }}
+
+{% if output_language %}
+**{{ "输出语言 / Output Language" if is_zh else "Output Language" }}:** {{ output_language }}
+{% endif %}
+
+{% if publication_date %}
+**{{ "发表日期 / Publication Date" if is_zh else "Publication Date" }}:** {{ publication_date }}
+{% endif %}
+
+{% if publication_venue %}
+**{{ "期刊/会议 / Publication Venue" if is_zh else "Publication Venue" }}:** {{ publication_venue }}
+{% endif %}
+
+{% if keywords %}
+**{{ "关键词 / Keywords" if is_zh else "Keywords" }}:** {{ keywords | join(", ") }}
+{% endif %}
+
+{% if abstract %}
+## {{ "摘要 / Abstract" if is_zh else "Abstract" }}
+
+{{ abstract }}
+{% endif %}
+
+{% if summary %}
+## {{ "总结 / Summary" if is_zh else "Summary" }}
+
+{{ summary }}
+{% endif %}

deepresearch_flow/paper/templates/eight_questions.md.j2

@@ -0,0 +1,49 @@
+# {{ paper_title }}
+
+{% set is_zh = output_language == "zh" %}
+
+**{{ "作者 / Authors" if is_zh else "Authors" }}:** {{ paper_authors | join(", ") }}
+
+{% if output_language %}
+**{{ "输出语言 / Output Language" if is_zh else "Output Language" }}:** {{ output_language }}
+{% endif %}
+
+{% if publication_date %}
+**{{ "发表日期 / Publication Date" if is_zh else "Publication Date" }}:** {{ publication_date }}
+{% endif %}
+
+{% if publication_venue %}
+**{{ "期刊/会议 / Publication Venue" if is_zh else "Publication Venue" }}:** {{ publication_venue }}
+{% endif %}
+
+## {{ "问题 1：论文试图解决什么问题？ / Question 1: What problem does the paper try to solve?" if is_zh else "Question 1: What problem does the paper try to solve?" }}
+
+{{ question1 }}
+
+## {{ "问题 2：有哪些相关研究？（最强基线/最相近思路/正交方向） / Question 2: What related research exists? (strongest baseline, closest approach, orthogonal directions)" if is_zh else "Question 2: What related research exists? (strongest baseline, closest approach, orthogonal directions)" }}
+
+{{ question2 }}
+
+## {{ "问题 3：论文如何解决这个问题？（流程图文字版） / Question 3: How does the paper solve the problem? (include a text flowchart)" if is_zh else "Question 3: How does the paper solve the problem? (include a text flowchart)" }}
+
+{{ question3 }}
+
+## {{ "问题 4：实验数据集与实现设置（数据集/场景、为何选择、验证哪些关键特性） / Question 4: Datasets and implementation settings (what datasets/scenarios, why chosen, what they validate)" if is_zh else "Question 4: Datasets and implementation settings (what datasets/scenarios, why chosen, what they validate)" }}
+
+{{ question4 }}
+
+## {{ "问题 5：指标与结果（指标、结果、各指标对应的创新点） / Question 5: Metrics and results (metrics, results, which innovations each metric reflects)" if is_zh else "Question 5: Metrics and results (metrics, results, which innovations each metric reflects)" }}
+
+{{ question5 }}
+
+## {{ "问题 6：有什么可以进一步探索的点？（理论/工程/数据/评测） / Question 6: What can be further explored? (theory/engineering/data/evaluation)" if is_zh else "Question 6: What can be further explored? (theory/engineering/data/evaluation)" }}
+
+{{ question6 }}
+
+## {{ "问题 7：现实场景如何体现/实现？（部署约束、算力、延迟、鲁棒性） / Question 7: How would this work in real-world scenarios? (deployment constraints, compute, latency, robustness)" if is_zh else "Question 7: How would this work in real-world scenarios? (deployment constraints, compute, latency, robustness)" }}
+
+{{ question7 }}
+
+## {{ "问题 8：总结主要内容（一段摘要 + 5 条要点 + 1 个核心公式/机制） / Question 8: Summarize the main content (one paragraph + 5 bullets + one core formula/mechanism)." if is_zh else "Question 8: Summarize the main content (one paragraph + 5 bullets + one core formula/mechanism)." }}
+
+{{ question8 }}

deepresearch_flow/paper/templates/three_pass.md.j2

@@ -0,0 +1,28 @@
+# {{ paper_title }}
+
+{% set is_zh = output_language == "zh" %}
+**{{ "作者 / Authors" if is_zh else "Authors" }}:** {{ paper_authors | join(", ") }}
+
+{% if output_language %}
+**{{ "输出语言 / Output Language" if is_zh else "Output Language" }}:** {{ output_language }}
+{% endif %}
+
+{% if publication_date %}
+**{{ "发表日期 / Publication Date" if is_zh else "Publication Date" }}:** {{ publication_date }}
+{% endif %}
+
+{% if publication_venue %}
+**{{ "期刊/会议 / Publication Venue" if is_zh else "Publication Venue" }}:** {{ publication_venue }}
+{% endif %}
+
+## {{ "步骤一：获取文章概要（标题/摘要/创新点/结构/图表/结论/参考/开源） / Step 1: Overview (title/abstract, innovations, structure, figures, conclusion, references, open source)" if is_zh else "Step 1: Overview (title/abstract, innovations, structure, figures, conclusion, references, open source)" }}
+
+{{ step1_summary }}
+
+## {{ "步骤二：理解内容（方法细节/实验/证据/相关工作缺陷） / Step 2: Content Comprehension (method details, experiments, evidence, related work gaps)" if is_zh else "Step 2: Content Comprehension (method details, experiments, evidence, related work gaps)" }}
+
+{{ step2_analysis }}
+
+## {{ "步骤三：深入研究和分析（重建/假设/失败模式/强弱点/未来工作） / Step 3: Deep Analysis (reconstruction, assumptions, failure modes, strengths/weaknesses, future work)" if is_zh else "Step 3: Deep Analysis (reconstruction, assumptions, failure modes, strengths/weaknesses, future work)" }}
+
+{{ step3_analysis }}

deepresearch_flow/paper/utils.py

@@ -0,0 +1,136 @@
+"""Utility helpers for paper extraction."""
+
+from __future__ import annotations
+
+import hashlib
+import json
+from pathlib import Path
+import re
+from typing import Iterable
+
+import json_repair
+
+
+def discover_markdown(
+    inputs: Iterable[str],
+    glob_pattern: str | None,
+    recursive: bool = True,
+) -> list[Path]:
+    files: set[Path] = set()
+    for raw in inputs:
+        path = Path(raw)
+        if path.is_file():
+            if path.suffix.lower() != ".md":
+                raise ValueError(f"Input file is not a markdown file: {path}")
+            files.add(path.resolve())
+            continue
+
+        if path.is_dir():
+            if glob_pattern:
+                matches = path.rglob(glob_pattern) if recursive else path.glob(glob_pattern)
+            else:
+                matches = path.rglob("*.md") if recursive else path.glob("*.md")
+            for match in matches:
+                if match.is_file():
+                    files.add(match.resolve())
+            continue
+
+        raise FileNotFoundError(f"Input path not found: {path}")
+
+    return sorted(files)
+
+
+def read_text(path: Path) -> str:
+    try:
+        return path.read_text(encoding="utf-8")
+    except UnicodeDecodeError:
+        return path.read_text(encoding="latin-1")
+
+
+def compute_source_hash(content: str) -> str:
+    return hashlib.sha256(content.encode("utf-8", errors="ignore")).hexdigest()
+
+
+def short_hash(value: str) -> str:
+    return hashlib.sha1(value.encode("utf-8", errors="ignore")).hexdigest()[:8]
+
+
+def stable_hash(value: str) -> str:
+    return hashlib.sha1(value.encode("utf-8", errors="ignore")).hexdigest()
+
+
+def truncate_content(
+    content: str, max_chars: int, strategy: str
+) -> tuple[str, dict[str, int | str] | None]:
+    if max_chars <= 0 or len(content) <= max_chars:
+        return content, None
+
+    if strategy == "head":
+        truncated = content[:max_chars]
+        meta = {
+            "strategy": "head",
+            "original_chars": len(content),
+            "kept_chars": max_chars,
+        }
+        return truncated, meta
+
+    if strategy == "head_tail":
+        head_len = max_chars // 2
+        tail_len = max_chars - head_len
+        truncated = content[:head_len] + "\n\n...\n\n" + content[-tail_len:]
+        meta = {
+            "strategy": "head_tail",
+            "original_chars": len(content),
+            "kept_chars": max_chars,
+        }
+        return truncated, meta
+
+    raise ValueError(f"Unknown truncate strategy: {strategy}")
+
+
+def estimate_tokens(char_count: int) -> int:
+    return max(1, char_count // 4)
+
+
+def extract_json_from_text(text: str) -> str:
+    text = text.strip()
+    if text.startswith("```"):
+        text = text.strip("`")
+    if text.startswith("{") and text.endswith("}"):
+        return text
+
+    match = re.search(r"\{[\s\S]*\}", text)
+    if match:
+        return match.group(0)
+
+    raise ValueError("No JSON object found in response")
+
+
+def parse_json(text: str) -> dict:
+    try:
+        return json.loads(text)
+    except json.JSONDecodeError:
+        extracted = extract_json_from_text(text)
+        try:
+            return json.loads(extracted)
+        except json.JSONDecodeError:
+            data = json_repair.loads(extracted, skip_json_loads=True)
+            if not isinstance(data, dict):
+                raise ValueError("Repaired JSON did not produce an object")
+            return data
+
+
+def split_output_name(path: Path) -> str:
+    if path.name.lower() == "output.md" and path.parent.name:
+        return path.parent.name
+    return path.stem
+
+
+def unique_split_name(base: str, used: set[str], source: str) -> str:
+    if base not in used:
+        used.add(base)
+        return base
+    suffix = short_hash(source)
+    candidate = f"{base}_{suffix}"
+    used.add(candidate)
+    return candidate

deepresearch_flow/paper/web/__init__.py

@@ -0,0 +1,2 @@
+"""Web viewer for paper databases."""
+