geojson simple parser

easyidp/data.py

@@ -600,6 +600,8 @@ def __init__(self, data_dir, test_out):
             self.labelme_warn  = data_dir / "json_test" / "labelme_warn_img.json"
             self.labelme_err   = data_dir / "json_test" / "for_read_json.json"
 
+            self.geojson_soy   = data_dir / "json_test" / "2023_soybean_field.geojson"
+
             if isinstance(test_out, str):
                 test_out = Path(test_out)
             self.out = test_out / "json_test"

easyidp/jsonfile.py

@@ -1,6 +1,12 @@
 import json
 import os
 import numpy as np
+import geojson
+import pyproj
+
+from rich.table import Table
+from rich.box import HORIZONTALS
+from rich import print
 
 class MyEncoder(json.JSONEncoder):
     # The original json package doesn't compatible to numpy object, add this compatible encoder to it.
@@ -54,6 +60,192 @@ def read_json(json_path):
             return data
     else:
         raise FileNotFoundError(f"Could not locate the given json file [{json_path}]")
+
+def _check_geojson_format(geojson_path):
+    if '.geojson' not in str(geojson_path):
+        raise TypeError(f"The input file format should be *.geojson, not [.{str(geojson_path).split('.')[-1]}]")
+
+    with open(geojson_path, 'r') as f:
+        geojson_data = geojson.load(f)
+
+    # Only the FeatureCollection geojson type is supported.
+    if geojson_data.type != 'FeatureCollection':
+        raise TypeError(f'A `FeatureCollection` type geojson is expected, current type is `{geojson_data.type}`')
+
+    if 'features' not in geojson_data.keys():
+        raise TypeError('geojson does not have features properties')
+
+    return geojson_data
+
+def read_geojson(geojson_path):
+    """Read geojson file to python dict
+
+    Parameters
+    ----------
+    geojson_path : str
+        The path to geojson file
+
+    Returns
+    -------
+    dict
+        .. code-block:: python
+
+            {
+                'crs': pyproj.CRS, 
+                'geometry': [ 
+                    {'coordiante': [[[...]]], 'type': '...'}, 
+                    ... 
+                ], 
+                'property': [ 
+                    {key1: v1, key2: v2},
+                    ... 
+                ]
+            }
+    
+    Example
+    -------
+    Data prepare:
+
+    .. code-block:: python
+
+        >>> import easyidp as idp
+        >>> test_data = idp.data.TestData()
+
+    Use this function:
+
+    .. code-block:: python
+
+        >>> out = idp.jsonfile.read_json(test_data.json.geojson_soy)
+        >>> out
+        {'crs': <Projected CRS: EPSG:6677>
+            Name: JGD2011 / Japan Plane Rectangular CS IX
+            Axis Info [cartesian]:
+            - X[north]: Northing (metre)
+            - Y[east]: Easting (metre)
+            Area of Use:
+            - name: Japan - onshore - Honshu - Tokyo-to.
+            - bounds: (138.4, 29.31, 141.11, 37.98)
+            Coordinate Operation:
+            - name: Japan Plane Rectangular CS zone IX
+            - method: Transverse Mercator
+            Datum: Japanese Geodetic Datum 2011
+            - Ellipsoid: GRS 1980
+            - Prime Meridian: Greenwich,
+         'geometry': [
+            { 
+                "coordinates": [[[-26384.952573, -28870.678514], ..., [-26384.952573, -28870.678514]]], 
+                "type": "Polygon" 
+            },
+            ...
+         ],
+         'property': [
+            {'FID': 65,
+             '試験区': 'SubBlk 2b',
+             'ID': 0,
+             '除草剤': '有',
+             'plotName': 'Enrei-10',
+             'lineNum': 1},
+            ...
+         ]
+        }
+
+    """
+    out_dict = {'crs': None, 'geometry': [], 'property':[]}
+
+    geojson_data = _check_geojson_format(geojson_path)
+
+    if 'crs' in geojson_data.keys():
+        # ensure it has correct format for CRS info
+        if  'properties' in geojson_data.crs and \
+            'name'       in geojson_data.crs['properties']:
+            out_dict['crs'] = pyproj.CRS.from_string(geojson_data.crs['properties']['name'])
+        else:
+            crs_template = {
+                "type": "name",
+                "properties": {
+                    "name": "EPSG:xxxx"
+                }
+            }
+            print(f'geojson does not have standard CRS properties like:\n{crs_template}'
+                  f'\nBut the following is obtained:\n{geojson_data.crs}')
+    else:
+        print(f'[json][geojson] geojson does not have CRS properties')
+
+    for f in geojson_data.features:
+        out_dict['geometry'].append(f['geometry'])
+        out_dict['property'].append(f['properties'])
+
+    return out_dict
+
+def show_geojson_fields(geojson_path):
+    """
+    Show geojson properties data, for better setting ``name_field`` of :py:obj:`read_geojson <easyidp.roi.ROI.read_geojson>`
+
+    Parameters
+    ----------
+    geojson_path : : str
+        the file path of \*.geojson
+
+    Example
+    -------
+
+    .. code-block:: python
+
+        >>> import easyidp as idp
+        >>> test_data = idp.data.TestData()
+
+        >>> idp.jsonfile.show_geojson_fields(test_data.json.geojson_soy)
+                          Properties of /Users/hwang/Library/Application                   
+             Support/easyidp.data/data_for_tests/json_test/2023_soybean_field.geojson      
+        ────────────────────────────────────────────────────────────────────────────────── 
+         [-1]   [0] FID   [1] 試験区   [2] ID   [3] 除草剤    [4] plotName    [5] lineNum  
+        ────────────────────────────────────────────────────────────────────────────────── 
+            0     65      SubBlk 2b      0          有          Enrei-10           1       
+            1     97      SubBlk 2b      0          有          Enrei-20           1       
+            2     147     SubBlk 2b      0          有        Nakasenri-10         1       
+          ...     ...        ...        ...        ...            ...             ...      
+          257     259       SB 0a        0                   Tachinagaha-10        3       
+          258      4        SB 0a        0                   Fukuyutaka-10         3       
+          259      1      SubBlk 2a      0          無          Enrei-20           1       
+        ──────────────────────────────────────────────────────────────────────────────────
+        
+    """
+    geojson_data = _check_geojson_format(geojson_path)
+
+    table = Table(title=f"Properties of {geojson_path}", box=HORIZONTALS)
+
+    table.add_column("[-1]", justify='right')
+    for i, c in enumerate(geojson_data.features[0]['properties'].keys()):
+        table.add_column(f"[{i}] {c}", justify='center')
+
+    row_num = len(geojson_data.features)
+    col_num = len(geojson_data.features[0]['properties'].keys())
+
+    if row_num >= 6:
+        show_idx_top = [0, 1, 2]
+        show_idx_btm = [-3, -2, -1]
+
+        for i in show_idx_top:
+            row = [str(i)] + [str(k) for k in geojson_data.features[i]['properties'].values()]
+            table.add_row(*row)
+
+        omit_row = ['...'] * (col_num + 1)
+        table.add_row(*omit_row)
+
+        for i in show_idx_btm:
+            row = [str(row_num + i)] + [str(k) for k in geojson_data.features[i]['properties'].values()]
+            table.add_row(*row)
+
+    else:
+        # print all without omit
+        show_idx = list(range(row_num))
+
+        for i in show_idx:
+            row = [str(i)] + list(geojson_data.features[i]['properties'].values())
+            table.add_row(*row)
+
+    print(table)
+
 
 def dict2json(data_dict, json_path, indent=None, encoding='utf-8'):
     """Save dict object to the same structure json file

tests/test_jsonfile.py

@@ -1,5 +1,6 @@
 import re
 import pytest
+import pyproj
 import numpy as np
 from pathlib import Path
 import easyidp as idp
@@ -38,3 +39,38 @@ def test_read_dict():
     out = idp.jsonfile.read_json(test_data.json.for_read_json)
 
     assert out == {"test": {"rua": [[12, 34], [45, 56]]}, "hua": [34, 34.567]}
+
+def test_read_geojson_wrong_format():
+    with pytest.raises(TypeError, match=re.escape("The input file format should be *.geojson, not [.json]")):
+        # input a *.json file
+        out = idp.jsonfile.read_geojson(test_data.json.labelme_demo)
+
+def test_read_geojson():
+    out = idp.jsonfile.read_geojson(test_data.json.geojson_soy)
+
+    assert out['crs'] == pyproj.CRS.from_epsg(6677)
+
+    assert len(out['geometry']) == len(out['property'])
+
+geojson_table_preview_unix = \
+"                   Properties of /Users/hwang/Library/Application                   \n"\
+"      Support/easyidp.data/data_for_tests/json_test/2023_soybean_field.geojson      \n"\
+" ────────────────────────────────────────────────────────────────────────────────── \n"\
+"  [-1]   [0] FID   [1] 試験区   [2] ID   [3] 除草剤    [4] plotName    [5] lineNum  \n"\
+" ────────────────────────────────────────────────────────────────────────────────── \n"\
+"     0     65      SubBlk 2b      0          有          Enrei-10           1       \n"\
+"     1     97      SubBlk 2b      0          有          Enrei-20           1       \n"\
+"     2     147     SubBlk 2b      0          有        Nakasenri-10         1       \n"\
+"   ...     ...        ...        ...        ...            ...             ...      \n"\
+"   257     259       SB 0a        0                   Tachinagaha-10        3       \n"\
+"   258      4        SB 0a        0                   Fukuyutaka-10         3       \n"\
+"   259      1      SubBlk 2a      0          無          Enrei-20           1       \n"\
+" ────────────────────────────────────────────────────────────────────────────────── \n"
+def test_show_geojson_fields(capfd):
+
+    idp.jsonfile.show_geojson_fields(test_data.json.geojson_soy)
+
+    out, err = capfd.readouterr()
+
+    # just ensure the function can run, no need to guarantee the same string outputs
+    assert "Properties of" in out