Merge pull request #171 from ToucanToco/prepare-v0.7.18

Prepare v0.7.18

CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## [0.7.18]
+
+### Changed
+
+- Added `text/xml` to supported XML mimetypes
+- Make `peakina.helpers.detect_type` return GEODATA if the stdlib returns a geojson-like mimetype
+
 ## [0.7.11] -  2022-04-05
 
 ### Changed

peakina/helpers.py

@@ -24,6 +24,7 @@
     excel_meta,
     read_csv,
     read_excel,
+    read_geo_data,
     read_json,
     read_xml,
 )
@@ -42,7 +43,7 @@ class TypeInfos(NamedTuple):
 
 
 # For files without MIME types, we make fake MIME types based on detected extension
-CUSTOM_MIMETYPES = {".parquet": "peakina/parquet"}
+CUSTOM_MIMETYPES = {".parquet": "peakina/parquet", ".geojson": "peakina/geo"}
 
 EXTRA_PEAKINA_READER_KWARGS = ["preview_offset", "preview_nrows"]
 
@@ -62,13 +63,17 @@ class TypeInfos(NamedTuple):
         ["encoding", "decimal"],
         excel_meta,
     ),
+    "geodata": TypeInfos(
+        ["peakina/geo"],
+        read_geo_data,
+    ),
     "json": TypeInfos(
         ["application/json"],
         read_json,
         ["filter"],  # this option comes from read_json, which @wraps(pd.read_json)
     ),
     "parquet": TypeInfos(["peakina/parquet"], pd.read_parquet),
-    "xml": TypeInfos(["application/xml"], read_xml),
+    "xml": TypeInfos(["application/xml", "text/xml"], read_xml),
 }
 
 
@@ -81,6 +86,7 @@ class TypeEnum(str, Enum):
     JSON = "json"
     PARQUET = "parquet"
     XML = "xml"
+    GEODATA = "geodata"
 
 
 def detect_type(filepath: str, is_regex: bool = False) -> Optional[TypeEnum]:
@@ -92,6 +98,9 @@ def detect_type(filepath: str, is_regex: bool = False) -> Optional[TypeEnum]:
         filepath = filepath.rstrip("$")
     mimetype, _ = mimetypes.guess_type(filepath)
 
+    if mimetype in ("application/geo+json", "application/vnd.geo+json"):
+        return TypeEnum.GEODATA
+
     # Fallback on custom MIME types
     if mimetype is None:
         _, fileext = os.path.splitext(filepath)

peakina/io/ftp/ftp_utils.py

@@ -42,9 +42,17 @@ def connect(  # type: ignore[override]
         self.port = port or 990
         self.timeout = timeout
 
-        self._sock = socket.create_connection((self.host, self.port), self.timeout)
-        self.af = self._sock.family
-        self.sock: ssl.SSLSocket = self.context.wrap_socket(self._sock, server_hostname=self.host)
+        def _setup_sock() -> socket.socket:
+            _sock = socket.create_connection((self.host, self.port), self.timeout)
+            self.af = _sock.family
+            return _sock
+
+        try:
+            self.sock = self.context.wrap_socket(_setup_sock(), server_hostname=self.host)
+        except ssl.SSLError:  # pragma: no cover
+            # in some cases we must fallback to:
+            self.sock = _setup_sock()
+
         self.file = self.sock.makefile("r")
         self.welcome = self.getresp()
         return self.welcome
@@ -57,7 +65,7 @@ def ntransfercmd(  # type: ignore[override]
         conn, size = ftplib.FTP.ntransfercmd(self, cmd, rest)
         if self._prot_p:  # type: ignore[attr-defined]
             conn = self.context.wrap_socket(
-                conn, server_hostname=self.host, session=self.sock.session
+                conn, server_hostname=self.host, session=self.sock.session  # type: ignore[union-attr]
             )  # this is the fix
         return conn, size
 
@@ -78,8 +86,17 @@ def ftps_client(params: ParseResult) -> Generator[Tuple[FTPS, str], None, None]:
     ftps = FTPS()
     try:
         ftps.connect(host=params.hostname or "", port=params.port, timeout=3)
-        ftps.prot_p()
-        ftps.login(user=params.username or "", passwd=params.password or "")
+        try:
+            ftps.prot_p()
+            ftps.login(user=params.username or "", passwd=params.password or "")
+        except Exception as e:
+            if "SSL/TLS required on the control channel" in str(e):
+                # This error means we should try the other way: first login, then prot_p:
+                ftps.login(user=params.username or "", passwd=params.password or "")
+                ftps.prot_p()
+            else:
+                raise
+
         yield ftps, params.path
 
     finally:

peakina/readers/__init__.py

@@ -1,5 +1,6 @@
 from .csv import csv_meta, read_csv
 from .excel import excel_meta, read_excel
+from .geodata import read_geo_data
 from .json import read_json
 from .xml import read_xml
 
@@ -14,4 +15,6 @@
     "read_json",
     # XML
     "read_xml",
+    # GEOJSON
+    "read_geo_data",
 )

peakina/readers/geodata.py

@@ -0,0 +1,16 @@
+from functools import wraps
+from typing import Any, Optional
+
+import geopandas as gpd
+
+
+@wraps(gpd.read_file)
+def read_geo_data(
+    path: str, preview_offset: int = 0, preview_nrows: Optional[int] = None, **kwargs: Any
+) -> gpd.GeoDataFrame:
+    if preview_nrows and not preview_offset:
+        return gpd.read_file(path, rows=preview_nrows, **kwargs)
+    else:
+        return gpd.read_file(
+            path, rows=slice(preview_offset, preview_nrows + 1 if preview_nrows else None), **kwargs
+        )