Project

General

Profile

« Previous | Next » 

Revision 297

Added by Aaron Marcuse-Kubitza about 13 years ago

bin/map: Factored out input processing code

View differences:

bin/map
78 78 
        '''Maps datasource to XML tree'''
79 79 
        doc1 = xml_dom.create_doc(out_label)
80 80 
        root = doc1.documentElement
81
82 
        def process(get_value, rows):
83 
            '''Processes input values
84 
            @param get_value f(in_, row):str
85 
            '''
86 
            for i, row in enumerate(rows):
87 
                if not (limit == None or i < limit): break
88 
                row_id = str(i)
89 
                for in_, out in mappings:
90 
                    value = metadata_value(in_)
91 
                    if value == None: value = get_value(in_, row)
92 
                    if value != None:
93 
                        try: xpath.put_obj(root, out, row_id, has_types, value)
94 
                        except Exception: traceback.print_exc()
95
81 96 
        if in_is_db:
82 97 
            assert in_is_xpaths
83 98
......
91 106
92 107 
            in_db = sql.connect(in_db_config)
93 108 
            in_pkeys = {}
94 
            for row_idx, row in enumerate(sql.rows(db_xml.get(in_db,
95 
                in_root_xml, in_pkeys, limit))):
96 
                row_id = str(row_idx)
109 
            def get_value(in_, row):
97 110 
                pkey, = row
98 
                for in_, out in mappings:
99 
                    value = metadata_value(in_)
100 
                    if value == None:
101 
                        in_ = in_.cloneNode(True) # don't modify orig value!
102 
                        xml_dom.set_id(xpath.get(in_, in_root), pkey)
103 
                        value = sql.value_or_none(db_xml.get(in_db, in_,
104 
                            in_pkeys))
105 
                    if value != None:
106 
                        try: xpath.put_obj(root, out, row_id, has_types,
107 
                            str(value))
108 
                        except Exception: traceback.print_exc()
111 
                in_ = in_.cloneNode(True) # don't modify orig value!
112 
                xml_dom.set_id(xpath.get(in_, in_root), pkey)
113 
                value = sql.value_or_none(db_xml.get(in_db, in_, in_pkeys))
114 
                if value != None: return str(value)
115 
                else: return None
116 
            process(get_value, sql.rows(db_xml.get(in_db, in_root_xml, in_pkeys,
117 
                limit)))
109 118 
            in_db.close()
110 119 
        elif in_is_xml:
111 
            row = xpath.get(doc0.documentElement, in_root)
112 
            for row_idx, row in enumerate(xml_dom.NodeElemIter(row.parentNode)):
113 
                if not (limit == None or row_idx < limit): break
114 
                row_id = str(row_idx)
115 
                for in_, out in mappings:
116 
                    value = metadata_value(in_)
117 
                    if value == None:
118 
                        node = xpath.get(row, in_)
119 
                        if node != None: value = xml_dom.value(node)
120 
                    if value != None:
121 
                        try: xpath.put_obj(root, out, row_id, has_types, value)
122 
                        except Exception: traceback.print_exc()
120 
            def get_value(in_, row):
121 
                node = xpath.get(row, in_)
122 
                if node != None: return xml_dom.value(node)
123 
                else: return None
124 
            row0 = xpath.get(doc0.documentElement, in_root)
125 
            process(get_value, xml_dom.NodeElemIter(row0.parentNode))
123 126 
        else: # input is CSV
124 127 
            map_ = dict(mappings)
125 128 
            reader = csv.reader(sys.stdin)
......
131 134 
                    try: mappings[i] = (col_idxs[in_], out)
132 135 
                    except KeyError: pass
133 136
134 
            for row_idx, row in enumerate(reader):
135 
                if not (limit == None or row_idx < limit): break
136 
                row_id = str(row_idx)
137 
                for in_, out in mappings:
138 
                    value = metadata_value(in_)
139 
                    if value == None:
140 
                        value = row[in_]
141 
                        if value == '': value = None
142 
                    if value != None:
143 
                        try: xpath.put_obj(root, out, row_id, has_types, value)
144 
                        except Exception: traceback.print_exc()
137 
            def get_value(in_, row):
138 
                value = row[in_]
139 
                if value != '': return value
140 
                else: return None
141 
            process(get_value, reader)
145 142 
        xml_func.process(root)
146 143 
        return doc1
147 144

Also available in: Unified diff