|
| 1 | +package fixedlength |
| 2 | + |
| 3 | +import ( |
| 4 | + "bufio" |
| 5 | + "errors" |
| 6 | + "io" |
| 7 | + "strings" |
| 8 | + "testing" |
| 9 | + |
| 10 | + "github.com/jf-tech/go-corelib/strs" |
| 11 | + "github.com/jf-tech/go-corelib/testlib" |
| 12 | + "github.com/stretchr/testify/assert" |
| 13 | + |
| 14 | + "github.com/jf-tech/omniparser/idr" |
| 15 | +) |
| 16 | + |
| 17 | +func TestIsErrInvalidEnvelope(t *testing.T) { |
| 18 | + assert.True(t, IsErrInvalidEnvelope(ErrInvalidEnvelope("test"))) |
| 19 | + assert.Equal(t, "test", ErrInvalidEnvelope("test").Error()) |
| 20 | + assert.False(t, IsErrInvalidEnvelope(errors.New("test"))) |
| 21 | +} |
| 22 | + |
| 23 | +func testReader(r io.Reader, decl *fileDecl) *reader { |
| 24 | + return &reader{ |
| 25 | + inputName: "test", |
| 26 | + r: bufio.NewReader(r), |
| 27 | + decl: decl, |
| 28 | + line: 1, |
| 29 | + } |
| 30 | +} |
| 31 | + |
| 32 | +func TestReadLine(t *testing.T) { |
| 33 | + r := testReader(strings.NewReader("abc\n\nefg\n \nxyz\n"), nil) |
| 34 | + assert.Equal(t, 1, r.line) |
| 35 | + |
| 36 | + line, err := r.readLine() |
| 37 | + assert.NoError(t, err) |
| 38 | + assert.Equal(t, []byte("abc"), line) |
| 39 | + assert.Equal(t, 2, r.line) |
| 40 | + |
| 41 | + // the second read will skip a truly empty line. |
| 42 | + line, err = r.readLine() |
| 43 | + assert.NoError(t, err) |
| 44 | + assert.Equal(t, []byte("efg"), line) |
| 45 | + assert.Equal(t, 4, r.line) |
| 46 | + |
| 47 | + // next line is not truly empty, it contains just spaces, we need to read it in. |
| 48 | + line, err = r.readLine() |
| 49 | + assert.NoError(t, err) |
| 50 | + assert.Equal(t, []byte(" "), line) |
| 51 | + assert.Equal(t, 5, r.line) |
| 52 | + |
| 53 | + line, err = r.readLine() |
| 54 | + assert.NoError(t, err) |
| 55 | + assert.Equal(t, []byte("xyz"), line) |
| 56 | + assert.Equal(t, 6, r.line) |
| 57 | + |
| 58 | + // io.EOF shouldn't bump up current line number. |
| 59 | + line, err = r.readLine() |
| 60 | + assert.Equal(t, io.EOF, err) |
| 61 | + assert.Equal(t, 6, r.line) |
| 62 | + |
| 63 | + // reading again should still return io.EOF and line number stays. |
| 64 | + line, err = r.readLine() |
| 65 | + assert.Equal(t, io.EOF, err) |
| 66 | + assert.Equal(t, 6, r.line) |
| 67 | + |
| 68 | + // Another scenario that io.Reader fails |
| 69 | + r = testReader(testlib.NewMockReadCloser("read error", nil), nil) |
| 70 | + assert.Equal(t, 1, r.line) |
| 71 | + line, err = r.readLine() |
| 72 | + assert.Error(t, err) |
| 73 | + assert.Equal(t, "read error", err.Error()) |
| 74 | + assert.Nil(t, line) |
| 75 | + // reading error (unless it's EOF) bumps current line number |
| 76 | + assert.Equal(t, 2, r.line) |
| 77 | +} |
| 78 | + |
| 79 | +func TestReadByRowsEnvelope_ByRowsDefault(t *testing.T) { |
| 80 | + // default by_rows = 1 |
| 81 | + r := testReader(strings.NewReader("abc\n\nefghijklmn\n \nxyz\n"), |
| 82 | + &fileDecl{Envelopes: []*envelopeDecl{{ |
| 83 | + Name: strs.StrPtr("env1"), |
| 84 | + Columns: []*columnDecl{ |
| 85 | + { |
| 86 | + Name: "col1", |
| 87 | + StartPos: 2, |
| 88 | + Length: 4, |
| 89 | + }, |
| 90 | + }, |
| 91 | + }}}) |
| 92 | + |
| 93 | + n, err := r.readByRowsEnvelope() |
| 94 | + assert.NoError(t, err) |
| 95 | + assert.Equal(t, `{"col1":"bc"}`, idr.JSONify2(n)) |
| 96 | + assert.Equal(t, 2, r.line) |
| 97 | + |
| 98 | + n, err = r.readByRowsEnvelope() |
| 99 | + assert.NoError(t, err) |
| 100 | + assert.Equal(t, `{"col1":"fghi"}`, idr.JSONify2(n)) |
| 101 | + assert.Equal(t, 4, r.line) |
| 102 | + |
| 103 | + n, err = r.readByRowsEnvelope() |
| 104 | + assert.NoError(t, err) |
| 105 | + assert.Equal(t, `{"col1":" "}`, idr.JSONify2(n)) |
| 106 | + assert.Equal(t, 5, r.line) |
| 107 | + |
| 108 | + n, err = r.readByRowsEnvelope() |
| 109 | + assert.NoError(t, err) |
| 110 | + assert.Equal(t, `{"col1":"yz"}`, idr.JSONify2(n)) |
| 111 | + assert.Equal(t, 6, r.line) |
| 112 | + |
| 113 | + n, err = r.readByRowsEnvelope() |
| 114 | + assert.Equal(t, io.EOF, err) |
| 115 | + assert.Nil(t, n) |
| 116 | +} |
| 117 | + |
| 118 | +func TestReadByRowsEnvelope_ByRowsNonDefault(t *testing.T) { |
| 119 | + r := testReader(strings.NewReader("abcdefg\n\nhijklmn\n \nabc012345\n"), |
| 120 | + &fileDecl{Envelopes: []*envelopeDecl{{ |
| 121 | + Name: strs.StrPtr("env1"), |
| 122 | + ByRows: testlib.IntPtr(3), |
| 123 | + Columns: []*columnDecl{ |
| 124 | + {Name: "col1", StartPos: 2, Length: 4, LinePattern: strs.StrPtr("^abc")}, |
| 125 | + {Name: "col2", StartPos: 2, Length: 4, LinePattern: strs.StrPtr("^hij")}, |
| 126 | + {Name: "col3", StartPos: 3, Length: 5, LinePattern: strs.StrPtr("^abc")}, |
| 127 | + }, |
| 128 | + }}}) |
| 129 | + |
| 130 | + n, err := r.readByRowsEnvelope() |
| 131 | + assert.NoError(t, err) |
| 132 | + assert.Equal(t, `{"col1":"bcde","col2":"ijkl","col3":"cdefg"}`, idr.JSONify2(n)) |
| 133 | + |
| 134 | + n, err = r.readByRowsEnvelope() |
| 135 | + assert.Error(t, err) |
| 136 | + assert.Equal(t, "input 'test' line 6: incomplete envelope, missing 2 row(s)", err.Error()) |
| 137 | + assert.Nil(t, n) |
| 138 | +} |
| 139 | + |
| 140 | +var ( |
| 141 | + benchReadByRowsEnvelopeInput = strings.Repeat( |
| 142 | + "abcdefghijklmnopqrstuvwxyz\n \n012345678901234567890123456789\n", 1000) |
| 143 | + benchReadByRowsEnvelopeDecl = &fileDecl{ |
| 144 | + Envelopes: []*envelopeDecl{ |
| 145 | + { |
| 146 | + Name: strs.StrPtr("env1"), |
| 147 | + ByRows: testlib.IntPtr(3), |
| 148 | + Columns: []*columnDecl{ |
| 149 | + {Name: "col1", StartPos: 2, Length: 10, LinePattern: strs.StrPtr("^abc")}, |
| 150 | + {Name: "col2", StartPos: 2, Length: 10, LinePattern: strs.StrPtr("^0123")}, |
| 151 | + {Name: "col3", StartPos: 12, Length: 19, LinePattern: strs.StrPtr("^abc")}, |
| 152 | + }, |
| 153 | + }, |
| 154 | + }, |
| 155 | + } |
| 156 | +) |
| 157 | + |
| 158 | +// BenchmarkReadByRowsEnvelope-8 624 1891740 ns/op 133140 B/op 9005 allocs/op |
| 159 | +func BenchmarkReadByRowsEnvelope(b *testing.B) { |
| 160 | + for i := 0; i < b.N; i++ { |
| 161 | + r := testReader(strings.NewReader(benchReadByRowsEnvelopeInput), benchReadByRowsEnvelopeDecl) |
| 162 | + for { |
| 163 | + n, err := r.readByRowsEnvelope() |
| 164 | + if err != nil { |
| 165 | + if err == io.EOF { |
| 166 | + break |
| 167 | + } |
| 168 | + b.FailNow() |
| 169 | + } |
| 170 | + idr.RemoveAndReleaseTree(n) |
| 171 | + } |
| 172 | + } |
| 173 | +} |
0 commit comments