LangSC 2.1.0__tar.gz → 2.2.2__tar.gz

langsc-2.2.2/LangSC/__version__.py

@@ -0,0 +1,7 @@
+# Single source of truth for the package version.
+# Keep __version__ a plain string literal: setup.cfg reads it statically via
+# `version = attr: LangSC.__version__.__version__`, so it must stay AST-parseable
+# (no computed expression) to avoid importing the package at build time.
+__version__ = "2.2.2"
+
+VERSION = tuple(int(x) for x in __version__.split("."))

{langsc-2.1.0 → langsc-2.2.2}/LangSC/bcc.py

@@ -24,12 +24,12 @@ lock = threading.Lock()
 class BCC:
     def __init__(self, dataPath="./data"):
         dataPath = dataPath.replace("\\", "/")
-        if dataPath[-1] == "/":
+        if len(dataPath) > 1 and dataPath.endswith("/"):
             dataPath = dataPath[:-1]
-        if dataPath.find("./") != 0:
-            dataPath = "./" + dataPath
-        if dataPath.find(".") != 0:
-            dataPath = "." + dataPath
+        # 相对路径规范为 "./" 前缀；绝对路径（/... 或 X:/...）原样保留
+        if not (os.path.isabs(dataPath) or (len(dataPath) >= 2 and dataPath[1] == ":")):
+            if not dataPath.startswith("./"):
+                dataPath = "./" + dataPath
 
         dll_name_bcc = ''
         self.g_IdxLog = "IdxLog_BCC.txt"

{langsc-2.1.0 → langsc-2.2.2}/LangSC/gpf.py

@@ -28,12 +28,12 @@ lock = threading.Lock()
 class GPF:
     def __init__(self, dataPath="./data"):
         dataPath = dataPath.replace("\\", "/")
-        if dataPath[-1] == "/":
+        if len(dataPath) > 1 and dataPath.endswith("/"):
             dataPath = dataPath[:-1]
-        if dataPath.find("./") != 0:
-            dataPath = "./" + dataPath
-        if dataPath.find(".") != 0:
-            dataPath = "." + dataPath
+        # 相对路径规范为 "./" 前缀；绝对路径（/... 或 X:/...）原样保留
+        if not (os.path.isabs(dataPath) or (len(dataPath) >= 2 and dataPath[1] == ":")):
+            if not dataPath.startswith("./"):
+                dataPath = "./" + dataPath
 
         dll_name_gpf = ''
         self.g_IdxLog = "IdxLog_GPF.txt"

{langsc-2.1.0 → langsc-2.2.2}/LangSC/jss.py

@@ -1,10 +1,23 @@
-"""JSS (JSON Structure Search) class wrapping jsslib.dll."""
+"""JSS (JSON Structure Search) class wrapping jsslib.dll.
 
+Besides local SQL search over jsslib.dll, the JSS class also exposes a set of
+``KG*`` methods for the remote **knowledge-graph** service. That service runs in
+the same process as the language-element search API (route prefix ``/kg/*``,
+Neo4j backend); the ``KG*`` methods talk to it over HTTP via ``requests`` and do
+NOT touch the local DLL. The default entry is the public gateway
+``https://cit.blcu.edu.cn/jss`` (use ``http://127.0.0.1:8100`` for a direct
+local service); override it with the ``kg_url`` constructor argument.
+"""
+
+import io
 import os
+import zipfile
 import platform
 import json
 from ctypes import *
 
+import requests
+
 from ._utils import (
     detect_file_encoding, get_idx_info, get_jss_file_info, is_same,
     get_jss_files, write_idx_log,
@@ -14,16 +27,23 @@ OS = platform.system()
 if OS == "Windows":
     import win32api
 
+# 知识图谱服务默认入口（公网 nginx）。直连本机服务用 http://127.0.0.1:8100
+DEFAULT_KG_BASE = "https://cit.blcu.edu.cn/jss"
+
 
 class JSS:
-    def __init__(self, dataPath, log_level=0, log_filename=''):
+    def __init__(self, dataPath, log_level=0, log_filename='', kg_url=DEFAULT_KG_BASE, kg_timeout=60):
+        # Knowledge-graph remote service (independent of the local DLL below)
+        self.kg_url = (kg_url or DEFAULT_KG_BASE).rstrip("/")
+        self.kg_timeout = kg_timeout
+
         dataPath = dataPath.replace("\\", "/")
-        if dataPath[-1] == "/":
+        if len(dataPath) > 1 and dataPath.endswith("/"):
             dataPath = dataPath[:-1]
-        if dataPath.find("./") != 0:
-            dataPath = "./" + dataPath
-        if dataPath.find(".") != 0:
-            dataPath = "." + dataPath
+        # 相对路径规范为 "./" 前缀；绝对路径（/... 或 X:/...）原样保留
+        if not (os.path.isabs(dataPath) or (len(dataPath) >= 2 and dataPath[1] == ":")):
+            if not dataPath.startswith("./"):
+                dataPath = "./" + dataPath
 
         dll_name_jss = ''
 
@@ -178,9 +198,10 @@ class JSS:
         else:
             seg_fullname = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'base.lex')
 
+        # JL_CreateTable 第 7 参数 pBuildToolInfo（当前引擎 API；旧版 6 参数 DLL 在 cdecl 下忽略多余实参，向后兼容）
         self.library_jss.JL_CreateTable.restype = c_int
-        self.library_jss.JL_CreateTable.argtypes = [c_char_p, c_char_p, c_char_p, c_char_p, c_int, c_char_p]
-        ret = self.library_jss.JL_CreateTable(cfg_filename.encode(), seg_fullname.encode(), dat_pathname.encode(), out_pathname.encode(), self.log_level, self.log_filename.encode())
+        self.library_jss.JL_CreateTable.argtypes = [c_char_p, c_char_p, c_char_p, c_char_p, c_int, c_char_p, c_char_p]
+        ret = self.library_jss.JL_CreateTable(cfg_filename.encode(), seg_fullname.encode(), dat_pathname.encode(), out_pathname.encode(), self.log_level, self.log_filename.encode(), b"LangSC")
         return ret
 
     def _load_table(self, filename, name=""):
@@ -214,3 +235,95 @@ class JSS:
                 self.library_jss.JL_Terminate(handle)
             self.handles = {}
             self.is_init = False
+
+    # ── Knowledge Graph (remote HTTP, /kg/* on self.kg_url) ──────────────────
+    # 这些方法走 HTTP（requests），不依赖本地 jsslib.dll，也不需要已加载的本地表。
+    # 后端为 Neo4j；服务不可用（未起 Neo4j / KG_ENABLED=0 等）时接口返回 HTTP 503。
+    # 节点统一结构 {id,label,type,graph_id,props?}，边 {id,source,target,type,label,graph_id,props?}。
+
+    def _kg_url(self, path):
+        return self.kg_url + "/" + path.lstrip("/")
+
+    def _kg_result(self, resp, ctx):
+        """Parse a /kg/* response; raise RuntimeError on any non-2xx (e.g. 400/404/503)."""
+        try:
+            parsed = resp.json()
+        except ValueError:
+            parsed = None
+        if not (200 <= resp.status_code < 300):
+            detail = parsed if parsed is not None else resp.text
+            raise RuntimeError("KG {} failed: HTTP {}: {}".format(ctx, resp.status_code, detail))
+        return parsed
+
+    @staticmethod
+    def _kg_zip_dir(dir_path):
+        """Pack a jss-kg/v1 package directory into zip bytes (files under an inner dir named after it)."""
+        base = os.path.basename(os.path.normpath(dir_path))
+        buf = io.BytesIO()
+        with zipfile.ZipFile(buf, "w", compression=zipfile.ZIP_DEFLATED) as zf:
+            for name in ("graph.json", "nodes.jsonl", "edges.jsonl", "nodes.csv", "edges.csv"):
+                full = os.path.join(dir_path, name)
+                if os.path.isfile(full):
+                    zf.write(full, arcname="{}/{}".format(base, name))
+        return buf.getvalue()
+
+    def KGImport(self, package, allow_reserved=False):
+        """Import a jss-kg/v1 package by upload (POST /kg/graphs/import, multipart).
+
+        ``package`` may be a .zip file path, a package directory
+        (containing graph.json + nodes.jsonl + edges.jsonl), or raw zip bytes.
+        Returns {graph_id, name, nodes, edges, warnings, replaced}.
+        """
+        if isinstance(package, (bytes, bytearray)):
+            zip_bytes = bytes(package)
+        elif os.path.isdir(package):
+            zip_bytes = self._kg_zip_dir(package)
+        else:
+            with open(package, "rb") as f:
+                zip_bytes = f.read()
+        files = {"file": ("graph.zip", zip_bytes, "application/zip")}
+        data = {"allow_reserved": "true"} if allow_reserved else {}
+        resp = requests.post(self._kg_url("kg/graphs/import"), files=files, data=data, timeout=self.kg_timeout)
+        return self._kg_result(resp, "import")
+
+    def KGImportPath(self, path, allow_reserved=False):
+        """Import a package already on the server under KG_IMPORT_ROOT, by relative path
+        (POST /kg/graphs/import_path, JSON). Returns the same shape as KGImport."""
+        payload = {"path": path, "allow_reserved": allow_reserved}
+        resp = requests.post(self._kg_url("kg/graphs/import_path"), json=payload, timeout=self.kg_timeout)
+        return self._kg_result(resp, "import_path")
+
+    def KGList(self):
+        """List imported graphs (GET /kg/graphs). Returns a list of graph-metadata dicts
+        [{graph_id, name, description, node_count, edge_count, default_hops, updated_at}], possibly empty."""
+        resp = requests.get(self._kg_url("kg/graphs"), timeout=self.kg_timeout)
+        parsed = self._kg_result(resp, "graphs")
+        if isinstance(parsed, dict):
+            return parsed.get("data", [])
+        return []
+
+    def KGDelete(self, graph_id):
+        """Delete a graph by id (DELETE /kg/graphs/{graph_id}). Returns {"deleted": graph_id}."""
+        resp = requests.delete(self._kg_url("kg/graphs/" + graph_id), timeout=self.kg_timeout)
+        return self._kg_result(resp, "delete")
+
+    def KGSubgraph(self, graph_id, q, hops=2, limit=200):
+        """Fetch a subgraph around seed ``q`` (POST /kg/subgraph).
+
+        Seed is matched by node id first, then by label. ``hops`` in {1,2,3};
+        ``limit`` in 1..2000. Returns {seed, nodes, edges}; on no match
+        {"nodes": [], "edges": [], "seed": None} with HTTP 200.
+        """
+        payload = {"graph_id": graph_id, "q": q, "hops": hops, "limit": limit}
+        resp = requests.post(self._kg_url("kg/subgraph"), json=payload, timeout=self.kg_timeout)
+        return self._kg_result(resp, "subgraph")
+
+    def KGExpand(self, graph_id, node_id, limit=50):
+        """Expand one hop around node ``node_id`` — by node id, NOT label (POST /kg/expand).
+
+        ``limit`` in 1..500. Returns {seed, nodes, edges}; {"nodes": [], "edges": [],
+        "seed": None} with HTTP 200 if the node does not exist.
+        """
+        payload = {"graph_id": graph_id, "id": node_id, "limit": limit}
+        resp = requests.post(self._kg_url("kg/expand"), json=payload, timeout=self.kg_timeout)
+        return self._kg_result(resp, "expand")

{langsc-2.1.0 → langsc-2.2.2/LangSC.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: LangSC
-Version: 2.1.0
+Version: 2.2.2
 Summary: Language Structure Computation - GPF, BCC, and JSS unified package
 Author: endongxun
 Author-email: xunendong@gmail.com

{langsc-2.1.0 → langsc-2.2.2}/LangSC.egg-info/SOURCES.txt

@@ -68,4 +68,5 @@ LangSC/Graph/librsvg-2-2.dll
 LangSC/Graph/libxml2.dll
 LangSC/Graph/ltdl.dll
 LangSC/Graph/zlib1.dll
-tests/test_gpf_threadsafe.py
+tests/test_gpf_threadsafe.py
+tests/test_win_smoke.py

{langsc-2.1.0/LangSC.egg-info → langsc-2.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: LangSC
-Version: 2.1.0
+Version: 2.2.2
 Summary: Language Structure Computation - GPF, BCC, and JSS unified package
 Author: endongxun
 Author-email: xunendong@gmail.com

{langsc-2.1.0 → langsc-2.2.2}/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = LangSC
-version = 2.1.0
+version = attr: LangSC.__version__.__version__
 description = Language Structure Computation - GPF, BCC, and JSS unified package
 long_description = file: README.md
 long_description_content_type = text/markdown

langsc-2.2.2/tests/test_win_smoke.py

@@ -0,0 +1,138 @@
+# -*- coding: utf-8 -*-
+"""LangSC 功能 + 集成冒烟测试（Windows 本地 DLL）。
+
+覆盖四个面：
+  GPF  — 分词/词性（本地 gpflib.dll + Segment.dat/idxPOS.dat）
+  JSS  — 由内联 JSONL 建表 + SQL 检索（本地 jsslib.dll + base.lex）
+  BCC  — 由内联语料建索引 + 频次检索（本地 bcclib.dll + Parser.lua）
+  KG   — 知识图谱远程集成（JSS.KG*，HTTP 到 cit.blcu.edu.cn；无网则跳过，不计失败）
+
+独立运行：python tests/test_win_smoke.py
+退出码 0 表示本地三引擎功能全通过（KG 为集成项，网络不可用时跳过）。
+"""
+import os
+import sys
+import json
+import shutil
+import tempfile
+
+try:
+    sys.stdout.reconfigure(encoding="utf-8")
+except Exception:
+    pass
+
+# 使 `import LangSC` 可用（tests/ 的上上级是包父目录）
+PKG_PARENT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.insert(0, PKG_PARENT)
+
+
+def section(title):
+    print("\n==================== {} ====================".format(title))
+
+
+def run_gpf():
+    from LangSC import GPF
+    tmp = tempfile.mkdtemp(prefix="lsc_gpf_")
+    g = GPF(tmp)
+    text = "我们到北京大学学习自然语言处理"
+    seg = json.loads(g.Parse(text, Structure="Segment"))
+    pos = json.loads(g.Parse(text, Structure="POS"))
+    print("Segment:", seg)
+    print("POS    :", pos)
+    assert isinstance(seg, list) and len(seg) >= 2, "分词结果应为多词列表"
+    assert isinstance(pos, list) and len(pos) >= 2, "词性结果应为多项列表"
+    assert any("/" in x for x in pos), "POS 项应形如 词/词性"
+    return True
+
+
+def run_jss():
+    # 注意：LangSC 的 dataPath 以相对路径使用（构造器会给非 "./" 开头的路径加前缀，
+    # 绝对路径会被破坏），故 chdir 到临时目录、用相对目录名。
+    from LangSC import JSS
+    base = tempfile.mkdtemp(prefix="lsc_jss_")
+    cwd = os.getcwd()
+    os.chdir(base)
+    try:
+        os.makedirs("data")
+        with open("data/doc.jsonl", "w", encoding="utf-8") as f:
+            f.write(json.dumps({"id": 1, "title": "学习汉语", "body": "我们在北京大学学习"}, ensure_ascii=False) + "\n")
+            f.write(json.dumps({"id": 2, "title": "自然语言处理", "body": "分词与词性标注"}, ensure_ascii=False) + "\n")
+        jss = JSS("data")                 # 自动建索引 + 加载
+        rows = jss.Run("SELECT id, title FROM doc WHERE id > 0;")
+        print("JSS rows:", rows)
+        assert isinstance(rows, list) and len(rows) == 2, "应检索到 2 条记录"
+        jss.Terminate()
+    finally:
+        os.chdir(cwd)
+    return True
+
+
+def run_bcc():
+    from LangSC import BCC
+    base = tempfile.mkdtemp(prefix="lsc_bcc_")
+    cwd = os.getcwd()
+    os.chdir(base)
+    try:
+        os.makedirs("corpus")
+        with open("corpus/text.txt", "w", encoding="utf-8") as f:
+            f.write("我们在北京大学学习自然语言处理。\n")
+            f.write("学习语言需要大量语料。\n")
+            f.write("自然语言处理很有意思。\n")
+        bcc = BCC("corpus")               # 自动分词建索引
+        ret = bcc.Run("学习", Command="Freq")
+        print("BCC Freq(学习):", str(ret)[:300])
+        env = json.loads(ret) if isinstance(ret, str) else ret
+        assert isinstance(env, dict) and env.get("type") == "Freq", "应返回 Freq 信封"
+    finally:
+        os.chdir(cwd)
+    return True
+
+
+def run_kg():
+    from LangSC import JSS
+    tmp = tempfile.mkdtemp(prefix="lsc_kg_")
+    os.makedirs(os.path.join(tmp, "d"))
+    jss = JSS(os.path.join(tmp, "d"))
+    graphs = jss.KGList()                  # 远程 HTTP
+    print("KG graphs:", [g.get("graph_id") for g in graphs] if isinstance(graphs, list) else graphs)
+    sg = jss.KGSubgraph("demo", "学习", hops=1, limit=20)
+    print("KG subgraph demo/学习: nodes=%d edges=%d" % (len(sg.get("nodes", [])), len(sg.get("edges", []))))
+    assert isinstance(graphs, list), "KGList 应返回列表"
+    return True
+
+
+def main():
+    # 默认只跑可靠、不会挂起的核心项：GPF（本地功能）+ KG（远程集成）。
+    # JSS/BCC 的建表/建索引走引擎 JL_CreateTable / BCC_IndexBCC，合成小数据下
+    # 目前会挂起或抛 C++ 异常（新旧 DLL 皆然，属引擎+数据问题，非构建回归），
+    # 故设为 opt-in：设 LSC_TEST_ENGINES=1 且提供真实数据目录时再启用。
+    cases = [
+        ("GPF 分词/词性", run_gpf, False),
+        ("KG 远程集成", run_kg, True),
+    ]
+    if os.environ.get("LSC_TEST_ENGINES") == "1":
+        cases.insert(1, ("JSS 建表/SQL", run_jss, False))
+        cases.insert(2, ("BCC 语料/频次", run_bcc, False))
+
+    results = []
+    for name, fn, integration in cases:
+        section(name)
+        try:
+            fn()
+            results.append((name, "PASS", integration))
+        except Exception as exc:
+            import traceback
+            traceback.print_exc()
+            results.append((name, "SKIP(net)" if integration else "FAIL", integration))
+
+    print("\n==================== SUMMARY ====================")
+    hard_fail = False
+    for name, status, integration in results:
+        print("  [{}] {}".format(status, name))
+        if status == "FAIL":
+            hard_fail = True
+    sys.exit(1 if hard_fail else 0)
+
+
+if __name__ == "__main__":
+    main()

langsc-2.1.0/LangSC/__version__.py

@@ -1,3 +0,0 @@
-VERSION = (2, 1, 0)
-
-__version__ = '.'.join(map(str, VERSION))