Readd Search API.

The suggestion api is still to be implemented.

Author: mgautierfr

libzim/libwrapper.h

@@ -28,6 +28,7 @@
 #include <zim/item.h>
 #include <zim/writer/item.h>
 #include <zim/writer/contentProvider.h>
+#include <zim/search.h>
 
 struct _object;
 typedef _object PyObject;
@@ -137,6 +138,7 @@ class WArchive : public Wrapper<zim::Archive>
     WArchive() = default;
     WArchive(const std::string& filename) : Wrapper(zim::Archive(filename)) {};
     WArchive(const zim::Archive& o) : Wrapper(o) {};
+    zim::Archive& operator*() const { return *mp_base; }
 
     FORWARD(WEntry, getEntryByPath)
     FORWARD(WEntry, getEntryByTitle)
@@ -167,6 +169,39 @@ class WArchive : public Wrapper<zim::Archive>
     FORWARD(bool, check)
 };
 
+class WSearchResultSet : public Wrapper<zim::SearchResultSet>
+{
+  public:
+    WSearchResultSet() = default;
+    WSearchResultSet(const zim::SearchResultSet& o) : Wrapper(o) {};
+
+
+    FORWARD(zim::SearchIterator, begin)
+    FORWARD(zim::SearchIterator, end)
+    FORWARD(int, size)
+};
+
+class WSearch : public Wrapper<zim::Search>
+{
+  public:
+    WSearch() = default;
+    WSearch(zim::Search&& s) : Wrapper(std::move(s)) {};
+
+    FORWARD(int, getEstimatedMatches)
+    FORWARD(WSearchResultSet, getResults)
+};
+
+class WSearcher : public Wrapper<zim::Searcher>
+{
+  public:
+    WSearcher() = default;
+    WSearcher(const WArchive& a) : Wrapper(zim::Searcher(*a)) {};
+    WSearcher(const zim::Searcher& o) : Wrapper(o) {};
+
+    FORWARD(void, setVerbose)
+    FORWARD(WSearch, search)
+};
+
 #undef FORWARD
 
 

libzim/searcher.py

@@ -0,0 +1,19 @@
+""" libzim reader module
+
+    - Archive to open and read ZIM files
+    - `Archive` gives access to all `Entry`
+    - `Entry` gives access to `Item` (content)
+
+    Usage:
+
+    with Archive(fpath) as zim:
+        entry = zim.get_entry_by_path(zim.main_entry.path)
+        print(f"Article {entry.title} at {entry.path} is "
+              f"{entry.get_item().content.nbytes}b")
+    """
+
+# flake8: noqa
+from .wrapper import Searcher, PyQuery as Query
+
+
+__all__ = ["Searcher", "Query"]

libzim/wrapper.pxd

@@ -76,6 +76,21 @@ cdef extern from "libwrapper.h":
     cdef cppclass WriterItemWrapper:
         WriterItemWrapper(PyObject* obj) except +
 
+cdef extern from "zim/search.h" namespace "zim":
+    cdef cppclass Query:
+        Query()
+        Query& setQuery(string query)
+        Query& setGeorange(float latitude, float longitude, float distance)
+
+
+cdef extern from "zim/search_iterator.h" namespace "zim":
+    cdef cppclass SearchIterator:
+        SearchIterator()
+        SearchIterator operator++()
+        bint operator==(SearchIterator)
+        bint operator!=(SearchIterator)
+        string getPath()
+        string getTitle()
 
 # Import the cpp wrappers.
 cdef extern from "libwrapper.h":
@@ -145,11 +160,17 @@ cdef extern from "libwrapper.h":
         bool hasChecksum() except +
         bool check() except +
 
-cdef extern from "zim/search_iterator.h" namespace "zim":
-    cdef cppclass SearchIterator:
-        SearchIterator()
-        SearchIterator operator++()
-        bint operator==(search_iterator)
-        bint operator!=(search_iterator)
-        string getPath()
-        string getTitle()
+    cdef cppclass WSearcher:
+        WSearcher()
+        WSearcher(const WArchive& archive) except +
+        setVerbose(bool verbose)
+        WSearch search(Query query) except +
+
+    cdef cppclass WSearch:
+        int getEstimatedMatches() except +
+        WSearchResultSet getResults(int start, int count) except +
+
+    cdef cppclass WSearchResultSet:
+        SearchIterator begin()
+        SearchIterator end()
+        int size()

libzim/wrapper.pyx

@@ -27,6 +27,7 @@ from typing import Generator
 from cython.operator import dereference, preincrement
 from cpython.ref cimport PyObject
 from cpython.buffer cimport PyBUF_WRITABLE
+from cython.operator import preincrement
 
 from libc.stdint cimport uint64_t
 from libcpp.string cimport string
@@ -647,92 +648,85 @@ cdef class PyArchive:
         except RuntimeError as e:
             raise KeyError(str(e))
 
-    def suggest(self, query: str, start: int = 0, end: int = 10) -> Generator[str, None, None]:
-        """ Paths of suggested entries in the archive from a title query -> Generator[str, None, None]
+    def __repr__(self) -> str:
+        return f"{self.__class__.__name__}(filename={self.filename})"
 
-            Parameters
-            ----------
-            query : str
-                Title query string
-            start : int
-                Iterator start (default 0)
-            end : end
-                Iterator end (default 10)
-            Returns
-            -------
-            Generator
-                Path of suggested entry """
-        # cdef wrapper.ZimSearch search = wrapper.ZimSearch(dereference(self.c_archive))
-        # search.set_suggestion_mode(True)
-        # search.set_query(query.encode('UTF-8'))
-        # search.set_range(start, end)
 
-        # cdef wrapper.search_iterator it = search.begin()
-        # while it != search.end():
-        #     yield it.get_path().decode('UTF-8')
-        #     preincrement(it)
+#########################
+#      Searcher         #
+#########################
 
-    def search(self, query: str, start: int = 0, end: int = 10) -> Generator[str, None, None]:
-        """ Paths of entries in the archive from a search query -> Generator[str, None, None]
+cdef class PyQuery:
+    cdef wrapper.Query c_query
 
-            Parameters
-            ----------
-            query : str
-                Query string
-            start : int
-                Iterator start (default 0)
-            end : end
-                Iterator end (default 10)
-            Returns
-            -------
-            Generator
-                Path of entry matching the search query """
+    def set_query(self, query: str):
+        self.c_query.setQuery(query.encode('utf8'))
+
+
+cdef class SearchResultSet:
+    cdef wrapper.WSearchResultSet c_resultset
+
+    @staticmethod
+    cdef from_resultset(wrapper.WSearchResultSet _resultset):
+        cdef SearchResultSet resultset = SearchResultSet()
+        resultset.c_resultset = move(_resultset)
+        return resultset
 
-        # cdef wrapper.ZimSearch search = wrapper.ZimSearch(dereference(self.c_archive))
-        # search.set_suggestion_mode(False)
-        # search.set_query(query.encode('UTF-8'))
-        # search.set_range(start, end)
+    def __iter__(self):
+        cdef wrapper.SearchIterator current = self.c_resultset.begin()
+        cdef wrapper.SearchIterator end = self.c_resultset.end()
+        while current != end:
+            yield current.getPath().decode('UTF-8')
+            preincrement(current)
 
-        # cdef wrapper.search_iterator it = search.begin()
-        # while it != search.end():
-        #     yield it.get_path().decode('UTF-8')
-        #     preincrement(it)
+cdef class Search:
+    cdef wrapper.WSearch c_search
 
-    def get_estimated_search_results_count(self, query: str) -> int:
-        """ Estimated number of search results for a query -> int
+    # Factory functions - Currently Cython can't use classmethods
+    @staticmethod
+    cdef from_search(wrapper.WSearch _search):
+        """ Creates a python ReadArticle from a C++ Article (zim::) -> ReadArticle
 
             Parameters
             ----------
-            query : str
-                Query string
+            _item : Item
+                A C++ Item
             Returns
-            -------
-            int
-                Estimated number of search results """
-        # cdef wrapper.ZimSearch search = wrapper.ZimSearch(dereference(self.c_archive))
-        # search.set_suggestion_mode(False)
-        # search.set_query(query.encode('UTF-8'))
-        # search.set_range(0, self.entry_count)
+            ------
+            Item
+                Casted item """
+        cdef Search search = Search()
+        search.c_search = move(_search)
+        return search
+
+    def getEstimatedMatches(self):
+        return self.c_search.getEstimatedMatches()
+
+    def getResults(self, start, count):
+        return SearchResultSet.from_resultset(move(self.c_search.getResults(start, count)))
+
+
+cdef class Searcher:
+    """ Zim Archive Searcher
+
+        Attributes
+        ----------
+        *c_archive : Searcher
+            a pointer to a C++ Searcher object
+    """
 
-        # return search.get_matches_estimated()
+    cdef wrapper.WSearcher c_searcher
 
-    def get_estimated_suggestions_results_count(self, query: str) -> int:
-        """ Estimated number of suggestions for a query -> int
+    def __cinit__(self, object archive: PyArchive):
+        """ Constructs an Archive from full zim file path
 
             Parameters
             ----------
-            query : str
-                Query string
-            Returns
-            -------
-            int
-                Estimated number of article suggestions """
-        # cdef wrapper.ZimSearch search = wrapper.ZimSearch(dereference(self.c_archive))
-        # search.set_suggestion_mode(True)
-        # search.set_query(query.encode('UTF-8'))
-        # search.set_range(0, self.entry_count)
+            filename : pathlib.Path
+                Full path to a zim file """
 
-        # return search.get_matches_estimated()
+        self.c_searcher = move(wrapper.WSearcher(archive.c_archive))
+
+    def search(self, object query: PyQuery):
+        return Search.from_search(move(self.c_searcher.search(query.c_query)))
 
-    def __repr__(self) -> str:
-        return f"{self.__class__.__name__}(filename={self.filename})"

tests/test_libzim_reader.py

@@ -10,6 +10,7 @@
 
 import libzim.writer
 from libzim.reader import Archive
+from libzim.searcher import Searcher, Query
 
 
 # expected data for tests ZIMs (see `all_zims`)
@@ -33,10 +34,10 @@
         "entry_count": 0,
         "all_entry_count": 2,
         "article_count": 0,
-        "suggestion_string": "",
+        "suggestion_string": None,
         "suggestion_count": 0,
         "suggestion_result": [],
-        "search_string": "",
+        "search_string": None,
         "search_count": 0,
         "search_result": [],
         "test_path": None,
@@ -124,9 +125,13 @@
         "suggestion_string": "favicon",
         "suggestion_count": 1,
         "suggestion_result": ["favicon.png"],
-        "search_string": "favicon",
-        "search_count": 1,
-        "search_result": ["favicon.png"],
+        "search_string": "main",
+        "search_count": 2,
+        "search_result": [
+            "Wikibooks.html",
+            "FreedomBox for Communities_Offline Wikipedia "
+            + "- Wikibooks, open books for an open world.html",
+        ],
         "test_path": "FreedomBox for Communities_Offline Wikipedia - Wikibooks, "
         "open books for an open world.html",
         "test_title": "FreedomBox for Communities/Offline Wikipedia - Wikibooks, "
@@ -398,15 +403,20 @@ def test_reader_suggest_search(
     assert zim.all_entry_count == all_entry_count
     assert zim.article_count == article_count
 
+    if search_string is not None:
+        query = Query()
+        query.set_query(search_string)
+        searcher = Searcher(zim)
+        search = searcher.search(query)
+        assert search.getEstimatedMatches() == search_count
+        assert list(search.getResults(0, search_count)) == search_result
 
-# TODO: restore [search-api]
+#TODO: restore suggestion search
 #     assert (
 #         zim.get_estimated_suggestions_results_count(suggestion_string)
 #         == suggestion_count
 #     )
 #     assert list(zim.suggest(suggestion_string)) == suggestion_result
-#     assert zim.get_estimated_search_results_count(search_string) == search_count
-#     assert list(zim.search(search_string)) == search_result
 
 
 @pytest.mark.parametrize(