Appendices

Format of .leo files

Here are the XML elements that may appear in Leo files:

<?xml>

Leo files start with the following line:

<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet>

This element is optional. For example:

<?xml-stylesheet ekr_stylesheet?>
<leo_file>

This element opens an element that contains the entire file. </leo_file> ends the file.

<leo_header>

This element specifies version information and other information that affects how Leo parses the file. For example:

<leo_header file_format="2"/>

The file_format attribute gives the ‘major’ format number. It is "2" for all newer versions of Leo.

<globals>, <preferences>, <find_panel_settings> and``<clone_windows>``

These elements are vestigial.
Modern versions of Leo ignore them when reading outlines and write empty elements when writing outlines.

<vnodes>

A single <vnodes> element contains nested <v> elements. <v> elements correspond to vnodes. The nesting of <v> elements indicates outline structure in the obvious way.

<v>

This element represents a single vnode and has the following form:

<v...><vh>sss</vh> (zero or more nested v elements) </v>

The <vh> element specifies the headline text. sss is the headline text encoded with the usual XML escapes. As shown above, a <v> element may contain nested <v> elements. Zero or more of the following attributes may appear in <v> elements:

t=name.timestamp.n
a="xxx"

The t="Tnnn" attribute specifies the <t> element associated with a <v> element. The a="xxx" attribute specifies vnode attributes. xxx denotes one or more upper-case letters whose meanings are as follows:

C       The vnode is a clone. (Not used in 4.x)
E       The vnode is expanded so its children are visible.
M       The vnode is marked.
T       The vnode is the top visible node.
V       The vnode is the current vnode.

For example, a="EM" specifies that the vnode is expanded and is marked.

<tnodes>

A single <tnodes> element contains a non-nested list of <t> elements.

<t>

This element represents the body text of the corresponding <v> element. It has this form:

<t tx="<gnx>">sss</t>

The tx attribute is required. The t attribute of <v> elements refer to this tx attribute. sss is the body text encoded with the usual XML escapes.

New in 4.0: Plugins and scripts may add attributes to <v> and <t> elements. See Writing plugins for details.

Format of external files

This section describes Leo’s sentinel lines, comment used in external files.

External files created by @file use gnx’s in @+node sentinels. Such gnx’s permanently and uniquely identify nodes. Gnx’s have the form:

id.yyyymmddhhmmss
id.yyyymmddhhmmss.n

The second form is used if two gnx’s would otherwise be identical.

  • id is a string unique to a developer, e.g., a git id.

  • yyyymmddhhmmss is the node’s creation date.

  • n is an integer.

Closing sentinels are required for section references and the @all and @others directives, collectively known as embedding constructs. Proof: These constructs do not terminate the node in which they appear. Without a closing sentinel there would be no way to know where the construct ended and the following lines of the enclosing node began.

Here are the sentinels used by modern versions of Leo, in alphabetical order.

@<<

Sentinels of the form @<<section_name>> represent section references.

If the reference does not end the line, the sentinel line ending the expansion is followed by the remainder of the reference line. This allows the Read code to recreate the reference line exactly.

@@

This sentinel represents any line starting with @ in body text except @<whitespace>, @doc and @others. Examples:

@@nocolor
@@pagewidth 80
@@tabwidth -4
@@c
@+all and -all

These sentinels mark the range of``@all`` directives.

@at and @doc

These sentinels mark the range of @doc parts.

@@delims

This sentinel marks the range of an @delims directive.
The range continues until the next @delims directive.

Adding, deleting or changing @@delim sentinels will destroy Leo’s ability to read the external file.

Mistakes in using the @delims directives have no effect on Leo, though such mistakes will thoroughly mess up a external file as far as compilers, HTML renderers, etc. are concerned.

@@section-delims

This sentinel represents the @section-delims directive.
Such must appear in the root @<file> node.

@+leo

This sentinel marks the start of any external file.
This sentinel has the form:

<opening_delim>@leo<closing_delim>

The read code uses single-line comments if <closing_delim> is empty.
The write code generates single-line comments if possible.

The @+leo sentinel contains other information. For example:

<opening_delim>@leo-ver=4-thin<closing_delim>
@-leo

This sentinel marks the end of the Leo file.
Nothing but whitespace should follow this directive.

@+node

This sentinel marks the start of a node:

@+node:gnx:<headline>

Nodes continue until the next @+node, at-all or at-others sentinel.

@+others and @-others

These sentinels mark the range of an @others directive.

@verbatim

This sentinel indicates that the next line of the external file is not a sentinel. This escape convention allows body text to contain lines that would otherwise be considered sentinel lines.

The Leonine way to refactor code

This paper explains how to use cff (clone-find-flattened) while refactoring code. I could not have completed the refactoring of Leo’s atFile write code without using continuous, extensive use of cff.

There are two key ideas:

  1. The clones produced by cff are short-term or medium-term data, easily created and easily dispensed with.

Such clones are valuable, but not precious. They will eventually be discarded.

  1. Unlike tags (or any other kind of non-Leonine data), the clones produced by cff can be reorganized.

This is the priceless, unique advantage of clones. You don’t understand clones if you don’t get this.

Example

  1. While refactoring, it is essential to see all actual uses of a symbol (method, or ivar, whatever).

The starting point is to use cff to find all potential uses of the symbol. If multiple files or classes use the symbol, you can use the suboutline-only option to limit the matches created by cff.

  1. After finding all potential uses of the symbol, you can reorganize the resulting clones as follows:

  • Delete nodes that are completely irrelevant.

  • Squirrel away likely-irrelevant nodes in a new organizer node.

  • Highlight the defining node, say by making it the preceding sibling of the cff node.

  • Leave all actual uses of the symbol where they are.

  1. You have now focused your attention on the nodes that will likely change.

You can now rerun the search only on those cloned nodes to see all instances of the symbol that might be changed. This is a crucial double check on proposed changes.

Summary

I highly recommend that all Leonine programmers use the approach just described when refactoring code.

Neither tags, nor filters, nor refactoring packages can emulate the Leonine way of refactoring.

Theory of operation: c.deletePositionsInList

The positions passed to p.deletePositionsInList only specify the desired changes; the only way to make those changes is to operate on vnodes!

Consider this outline, containing no clones:

+ ROOT
  - A
  - B

The fundamental problem is this. If we delete node A, the index of node B in ROOT.children will change. This problem has (almost) nothing to do with clones or positions.

Proof that c.deletePositionsInList is correct

Clearly, deleting a positions ultimately means changing vnodes, specifically, v.parents and v.children for v in some set of vnodes. We must show that the particular set of changed vnodes in the new code is the correct set of vnodes, and that the changes made to that set are correct. This is far from obvious at first glance.

The new code

Here is Vitalije’s version of c.deletePositionsInList, without the docstring:

def deletePositionsInList(self, aList):
    c = self

    # Ensure all positions are valid.
    aList = [p for p in aList if c.positionExists(p)]
    if not aList:
        return

    def p2link(p):
        parent_v = p.stack[-1][0] if p.stack else c.hiddenRootNode
        return p._childIndex, parent_v

    links_to_be_cut = sorted(set(map(p2link, aList)), key=lambda x:-x[0])

    # The main loop.
    for i, v in links_to_be_cut:
        child_v = v.children.pop(i)
        child_v.parents.remove(v)

To begin the proof, note that the main loop has the expected “form”. That is, when we delete a node, we will:

  1. Delete child_v = v.children[i] for some v and i.

  2. Delete child_v from its parents array.

v.parents is an unordered array, so child_v.parents.remove(v) is all that is needed. This line might crash if one of the positions is invalid. It will also crash if v is not present in child_v.parents. I’ll discuss this later.

To complete the proof, we must show that the main loop deletes all and only those vnodes implied by the positions in aList. We can tentatively assume that the main loop will work properly if that data in links_to_be_cut is correct, but there are some subtleties lurking here which I’ll discuss later.

The following code creates links_to_be_cut:

def p2link(p):
    parent_v = p.stack[-1][0] if p.stack else c.hiddenRootNode
    return p._childIndex, parent_v

links_to_be_cut = sorted(set(map(p2link, aList)), key=lambda x:-x[0])

This is elegant (compressed) code. Let’s examine it carefully…

p2link produces tuples (childIndex, parent_v). These become bound to i, v in the main loop:

# The main loop.
for i, v in links_to_be_cut:
    child_v = v.children.pop(i)
    child_v.parents.remove(v)

To make this work, parent_v must be the vnode whose i’th child should be deleted:

parent_v = p.stack[-1][0] if p.stack else c.hiddenRootNode

  • p.stack entries have the form (v, childIndex), so p.stack[-1] is the position p’s immediate parent vnode, if p has a parent. Otherwise, c.hiddenRootNode is the proper parent vnode.

  • p._childIndex is also the correct value for i.

We have just proved the following:

Deleting position p means executing the main loop for (i, v) returned by p2link(p)

Sorting and filtering

The following line filters and sorts the results produced by p2link:

links_to_be_cut = sorted(set(map(p2link, aList)), key=lambda x:-x[0]

Let’s break this into parts, from “inside” to “outside”

  • map(p2link, aList) applies p2link to every position in aList. The result is a list of tuples (i, v). This list may contain duplicates.

  • set(map(p2link, aList) removes those duplicates. Let theSet be set(map(p2link, aList).

  • Finally, sorted(theSet, key=lambda x:-x[0]) creates a generator equivalent to a sorted list.

The generator will deliver the tuples in descending order of i, when several tuples have the same vnode v. The generator does not order tuples on v, and there is no need to do so.

It’s easy to understand the intent of this sort order. The main loop contains this line:

child_v = v.children.pop(i)

The sort order means that pop(i) happens before pop(j) if i > j. This condition ensures that the precomputed values of i won’t change in the main loop.

Completing the proof

The algorithm appears sound. The tuples (i, v) computed from p2link(p) correctly describe the work to be done. Filtering ensures that the work is done once. Sorting ensures that child indices i will not change during the main loop.

The tuples are sorted on i, but not v. aList can contain positions in any order, so we know only that for any particular vnode v the main loop will handle tuples (i1, v), (i2, v), ... in the correct order.

There are two final questions to consider:

First, does the order in which the main loop handles vnodes matter?

No. The vnode data are independent. The main loop works regardless of whether a vnode has already been unlinked.

Second, could the following lines ever fail?

child_v = v.children.pop(i)
child_v.parents.remove(v)

No. Sorting and filtering ensure that tuples are unique. v is guaranteed to be in child_v.parents because v.children[i] must exist.

This concludes the proof.

Discussion

The algorithm is sound because the tuples (i, v) contain no duplicates. For any particular v, the main loop will process the tuples in descending order of i. This resolves the question of whether deleting already deleted positions is valid.

Leo’s Colorizer: Theory of Operation

Are you sure you want to read this appendix? It is intended only for Leo’s core developers.

This appendix contains the theory of operation for the JEditColorizer (jedit) class in leoColorizer.py. Unless qualified, all methods are members of the jedit class.

Executive summary

Use the --trace=coloring to see the colorizer in action.

The jedit class collaborates with the qsh, a singleton instance of the QSyntaxHighlighter class.

This collaboration allows the qsh to minimize the calls to jedit.redraw. The qsh calls redraw(s) only if line s could possibly need recoloring.

Leo’s core must never call redraw! These calls must happen automatically. Calling redraw from Leo’s core could cause hard-to-find bugs!

States allow the jedit class to handle constructs like strings or comments that continue from one line to another.

Mode files, python files in the leo/modes directory, tell the jedit class how to colorize specific languages.

Pattern matchers do the actual syntax coloring.

Components of the jedit class

The JEditColorizer class consists of drivers, pattern matchers and various helper utilities.

Drivers

The redraw and mainLoop methods are the interface between the qsh and the methods of the jedit class.

qsh.rehighlight calls redraw(s) to colorize a single line s. After initing the initial state, redraw calls mainLoop.

Leo’s core must never call redraw! Doing so could cause hard-to-find bugs.

redraw(s) calls mainLoop(s) to do the actual syntax coloring.

mainLoop(s) calls rule functions (defined in mode files).

Rule functions delegate their work to exactly one pattern matcher (described next). In other words, pattern matchers are callbacks.

Pattern matchers

Pattern matchers form the bulk of the jedit class. Each pattern matcher does the following:

  • Examines the string s[i:], where s is the argument to mainLoop(s).

  • Updates the ending state for successful matches.

  • Returns an integer return code:

    -1: complete failure: mainLoop returns
 0: partial failure: mainLoop tries to match other rules
>0: success: mainLoop increments i by n and tries to match other rules

The big challenge

The qsh can (and will) call recolor(s) where s is out-of-sequence with the preceding call. For example, the following snippet is typical in @language jupytext nodes:

# %% (like @language python)

def spam():

# %% [markdown] (like @language md)

# Section 1
It was the best of times; it was the worst of times.

The user can click (or cut or paste) anywhere in this text. The qsh will then call recolor(s) starting with the first changed line.

There is no necessary relationship between s and the line that recolor last saw!

About states and restarters

States allow the colorizer to handle out-of-order calls to recolor.

The colorizer collaborates the qsh in only one way, namely by calling self.highlighter.setCurrentBlockState(n), where n is an integer state number.

Internally, the colorizer uses string states. The colorizer’s state utilities handle the conversion between integer and string representations.

States represent, for a particular line:

  • Which @language directive is in effect.

  • Which coloring directive (@killcolor, @color, @nocolor, etc.) is in effect.

  • Whether the coloring for the line continues to the next line. For example, Python docstrings may span multiple lines.

A fine point: The state utilities ensure that all integer state numbers are unique and refer to the proper language in effect. For example, if a node contains multiple @language directives, the default states of the two languages must have different state numbers!

The colorizer keeps track of states as follows:

recolor first computes the line’s initial state from the ending state of the previous line. By default, the initial state becomes the ending state of current state.

Pattern matchers usually leave the ending state unchanged.

If a pattern matcher matches an @language or coloring directive, the matcher creates (indirectly, via the state utilities) a new state representing the directive. This is the easy case.

If a pattern matcher matches syntax that continues to the next line, the pattern matcher creates a restarter matcher by calling setRestart. The restarter matchers are essentially closures, bindings of the pattern matcher and all of its kwargs. The details are hairy, but the coding pattern is the same for all cases. See the source code for details.

Mode files

The jEdit editor drives its syntax colorer using .xml files. Leo uses .py files instead.

Long ago, Leo’s jEdit2Py script created the python mode files in the leo/modes directory.

Roughly speaking, there is one mode file for every language that Leo can colorize. Over the years, Leo’s devs have modified several of these mode files by hand.

init sets jedit.rulesDict from the mode file for the language in effect. Keys are lead-in characters; values are lists of rule functions (rules for short). Each mode file defines its own rulesDict and rules.

mainLoop(s) uses the rulesDict to retrieve a list of zero or more rules to be used when scanning s. It’s that simple.

Legacy mode files could collaborate using the delegate keyword, but that scheme was (and is) clumsy.

Leo 6.8.3 added leo/modes/jupytext.py, a new mode file for @language jupytext. This mode file introduced a new style of collaboration. The mode files for the jupytext, python and markdown languages collaborate directly using modifications of existing rules in python.py and md.py.

leoAst.py

The classes in leoAst.py unify python’s token-based and ast-based worlds by creating two-way links between tokens in the token list and ast nodes in the parse tree.

leoAst.py is part of Leo, and can be used completely independently of Leo.

You must use Leo to see the intended outline structure of the code. Without Leo, you will see special sentinel comments that create Leo’s outline structure. These comments have the form:

#@<comment-kind>:<user-id>.<timestamp>.<number>: <outline-level> <headline>

If you have any trouble installing or using this code, please help for help on Leo’s forum

Running leoAst.py

Running leoAst.py from the command line

leoAst.py is designed to be run from the command line:

usage:
    leoAst.py --help
    leoAst.py [--fstringify | --fstringify-diff | --orange | --orange-diff] PATHS
    leoAst.py --py-cov [ARGS]
    leoAst.py --pytest [ARGS]
    leoAst.py --unittest [ARGS]

examples:
    --py-cov "-f TestOrange"
    --pytest "-f TestOrange"
    --unittest TestOrange

positional arguments:
  PATHS              directory or list of files

optional arguments:
  -h, --help         show this help message and exit
  --fstringify       leonine fstringify
  --fstringify-diff  show fstringify diff
  --orange           leonine Black
  --orange-diff      show orange diff
  --py-cov           run pytest --cov on leoAst.py
  --pytest           run pytest on leoAst.py
  --unittest         run unittest on leoAst.py

Running the code python programmatically

To access the code, do one of the following:

code-block:

import leoAst
import leo.core.leoAst as leoAst

You can then run the fstringify commands as follows:

code-block:

changed = leoAst.Fstringify().fstringify_file(filename)
changed = leoAst.Fstringify().fstringify_diff_files(filename)

Running unit tests and coverage tests programmatically

The following runs all unit tests for leoAst.py:

python -m leo.core.leoAst

The following runs coverage tests:

pytest -x --cov-report html --cov-report term-missing --cov=leo.core.leoAst leo/core/leoAst.py

TokenOrder classes: Theory of operation

This is the Theory of Operation for the TokenOrderGenerator (TOG) class and related classes.

Token-order classes

The TokenOrderGenerator (TOG) class injects two-way links between all tokens and the corresponding ast nodes. The TOG class also injects parent/child links into all ast nodes.

The TOG class defines visitors that visit ast nodes in token order, the traversal order that corresponds to the order of the tokens produced by python’s tokenizer module. The only way to ensure this correspondence is to use separate visitors for all ast nodes. All visitors are straightforward generators.

TOG visitors eventually call TOG.sync_token, which checks that that tokens are, in fact, visited in the correct order. TOG.sync_token is an ever-present unit test.

The TokenOrderTraversal (TOT) class uses the parent/child links created by the TOG class. TOT.traverse contains a single for-loop that calls all nodes of the parse tree in token order. This loop is extremely fast. Using the TOT class, client code can easily modify the token list or parse tree as desired.

Other classes

The Token class represents one token, created by tokenize.tokenize.

The Fstringify class is an re-implementation of the external fstringify project using the TOG class.

The Orange class is a re-implementation of the black project.
The name “Orange” is a play on words: “Orange is the new black”.

The AstDumper class provides an extensive set of tools for examining token lists, parse trees, and the links between them.

The BaseTest class provides common infrastructure for all other test classes. Important: BaseTest.make_data is, all by itself, a very strong unit test.

Significant vs insignificant tokens

The distinction between significant and insignificant tokens is crucial. Visitors call TOG.gen_token, TOG.gen_op, etc. only for significant tokens. The is_significant and is_significant_token functions define which tokens are significant.

Visitors can’t know, just by looking at the parse tree, whether the input contains insignificant tokens. For example, the source tokens might contain non-essential parentheses, or optional trailing commas, or whether two statements are separated by a semicolon.

Helping TOG visitors

The TOG.do_If visitor calls TOG.find_next_significant_token to determine whether TOG.do_If should generate an “if” or an “elif” token. This help is essential, because the following two source files generate identical parse trees!

if 1:            if 1:
    pass             pass
else:            elif 2:
    if 2:            pass
        pass

Similarly, the TOG.do_Str and TOG.do_JoinedStr visitors call TOG.get_concatenated_string_tokens to handle one ore more concatenated string tokens.

Finally, TOG.do_slice calls TOG.find_next_significant_token to determine whether a slice without a step contains an optional colon.

TOG.find_next_significant_token and TOG.get_concatenated_string_tokens are crucial inventions. TOG class would not be possible without them.

Syncing tokens

TOG.px is an index into the token list. It is either -1, or it points at the previous significant token. Note: TOG.find_next_significant_token and TOG.get_concatenated_string_tokens use TOG.px, but never change TOG.px.

TOG.sync_token(self, kind, val) associates tokens with ast nodes as follows:

  1. If (kind, val) denote an insignificant token, TOG.sync_token does nothing.

  2. Otherwise, (kind, val) denotes a significant token. TOG.sync_token associates that token, plus all previous insignificant tokens with self.node, the ast node presently being visited.

In addition, if (kind, val) denotes a significant token, TOG.sync_token checks that the next significant token in the token list has the expected kind and value. This is done as follows:

  • TOG.sync_token advances TOG.px to point at the next significant token, call it T.

  • TOG.raises AssignLinksError if (T.kind, T.value) != (kind, val)

To summarize token syncing:

  • The TOG.px index tracks the last-seen significant token.

  • TOG.px advances monotonically through the token list.

  • TOG.find_next_significant_token and TOG.get_concatenated_string_tokens scan forward through the token list using a private copy of TOG.px. These methods never change TOG.px itself.

  • This token-syncing machinery is the simplest thing that could possibly work. It is also the fastest thing that could possibly work.

Figures of merit

Simplicity:

  • The distinction between significant and insignificant tokens makes token-order traversals possible. This distinction drastically simplifies TOG visitors. They never have to generate insignificant tokens!

  • TOG.find_next_significant_token and TOG.get_concatenated_string_tokens() use TOG.px to look ahead in the token list.

  • It took a long time to realize that the parse tree needs help from the token list, not the other way around!

Speed: The TOG creates links between tokens and ast nodes in roughly the time taken by python’s tokenize.tokenize and ast.parse library methods. The TOT class traverses trees annotated with parent/child links even more quickly.

TOG class avoids both ast.fix_missing_locations and ast.get_source_segment, which are too slow to be useful.

Memory: The TOG class makes no significant demand on python’s resources:

  • Generators add nothing to python’s call stack.

  • The only variable-length data created by the TOG is TOG.node_stack. This stack resides in python’s heap, so its length is unimportant. In the worst case, it might contain a few thousand entries.

  • The TOT uses no variable-length data whatever.

Maintaining leoAst.py

New ast nodes are sometimes required to support new language features, especially language features that require new syntax or keywords. Python has added new nodes fairly often in the past. New nodes may be added in future. When that happens, the following changes will be needed to leoAst.py:

  • Add a visitor for the new node.

  • Add one or more unit tests that fully cover the new visitor.

The test_visitors_exist unit test checks that visitors exist for all ast nodes defined by to a particular version of python.

See Leo issue #1440 for notes relating to the code.

Unicode reference

Leo uses unicode internally for all strings.

  1. Leo converts headline and body text to unicode when reading .leo files and external files. Both .leo files and external files may specify their encoding. The default is utf-8. If the encoding used in a external file is not “utf-8” it is represented in the @+leo sentinel line. For example:

        #@+leo-encoding=iso-8859-1.

The utf-8 encoding is a "lossless" encoding (it can represent all
unicode code points), so converting to and from utf-8 plain
strings will never cause a problem. When reading or writing a
character not in a "lossy" encoding, Leo converts such characters
to '?' and issues a warning.

  2. When writing .leo files and external files Leo uses the same encoding used to read the file, again with utf-8 used as a default.

  3. leoSettings.leo contains the following Unicode settings, with the defaults as shown:

        default_derived_file_encoding = UTF-8
    new_leo_file_encoding = UTF-8

These control the default encodings used when writing external
files and .leo files. Changing the new_leo_file_encoding setting
is not recommended. See the comments in leoSettings.leo. You may
set default_derived_file_encoding to anything that makes sense for
you.

  4. The @encoding directive specifies the encoding used in a external file. You can’t mix encodings in a single external file.

Valid URL’s

Leo checks that the URL is valid before attempting to open it. A valid URL is:

  • 3 or more lowercase alphas

  • followed by one :

  • followed by one or more of:

  • $%&'()*+,-./0-9:=?@A-Z_a-z{}~

  • followed by one of: $%&'()*+/0-9:=?@A-Z_a-z}~

That is, a comma, hyphen and open curly brace may not be the last character.

URL’s in Leo should contain no spaces: use %20 to indicate spaces.

You may use any type of URL that your browser supports: http, mailto, ftp, file, etc.

The Mulder/Ream update algorithm

This appendix documents the Mulder/Ream update algorithm in detail, with an informal proof of its correctness.

Prior to Leo 5.1, Leo used Bernhard Mulder’s original algorithm to read @shadow files. Starting with Leo 5.1, Leo uses this algorithm to read both @clean and @shadow files. Conceptually, both algorithms work as described in the next section.

In February 2015 EKR realized that the @shadow algorithm could be used to update @clean (@nosent) files. Simplifying the algorithm instantly became a top priority. The new code emerged several days later, made possible by the x.sentinels array. It is an important milestone in Leo’s history.

What the algorithm does

For simplicity, this discussion will assume that we are updating an external file, x, created with @clean x. The update algorithm works exactly the same way with @shadow trees.

The algorithm works with any kind of text file. The algorithm uses only difflib. It knows nothing about the text or its meaning. No parsing is ever done.

Suppose file x has been changed outside of Leo. When Leo reads x it does the following:

  1. Recreates the old version of x without sentinels by writing the @clean x outline into a string, as if it were writing the @clean x outline again.

  2. Recreates all the lines of x with sentinels by writing the @clean x outline into a string, as if it was writing an @file node! Let’s call these lines the old sentinels lines.

  3. Uses difflib.SequenceMatcher to create a set of diffs between the old and new versions of x without sentinels.

    Terminology: the diffs tell how to change file a into file b. The actual code uses this terminology: a is set of lines in the old version of x, b is the set of lines in the new version of x.

  4. Creates a set of lines, the new sentinels lines using the old sentinels lines, the a and b lines and the diffs.

    This is the magic. Bernhard Mulder’s genius was conceiving that a three-way merge of lines could produce the new outline, with sentinels. The code is in x.propagate_changed_lines and its helpers.

  5. Replaces the @clean tree with the new tree created by reading the new sentinels lines with the @file read logic.

Important: The update algorithm never changes sentinels. It never inserts or deletes nodes. The user is responsible for creating nodes to hold new lines, or for deleting nodes that become empty as the result of deleting lines.

Guesses don’t matter

There are several boundary cases that the update algorithm can not resolve. For example, if a line is inserted between nodes, the algorithm can not determine whether the line should be inserted at the end of one node or the start of the next node. Let us call such lines ambiguous lines.

The algorithm guesses that ambiguous lines belongs at the end of a node rather than at the start of the next node. This is usually what is wanted–we usually insert lines at the end of a node.

Happily, guesses don’t matter, for the following reasons:

  1. The external file that results from writing the @clean x tree will be the same as the updated external file no matter where ambiguous lines are placed. In other words, the update algorithm is sound.

  2. Leo reports nodes that were changed when reading any external file. The user can review changes to @clean and @file trees in the same way.

  3. The user can permanently correct any mistaken guess. Guesses only happen for newly inserted or changed lines. Moving an ambiguous line to the following node will not change the external file. As a result, the next time Leo reads the file the line will be placed in the correct node!

This proves that @shadow and @clean are easy and safe to use. The remaining sections of this document discuss code-level details.

Background of the code

The algorithm depends on three simple, guaranteed, properties of SequenceMatcher.opcodes. See https://docs.python.org/2/library/difflib.html#sequencematcher-examples

Fact 1: The opcodes tell how to turn x.a (a list of lines) into x.b (another list of lines).

The code uses the a and b terminology. It’s concise and easy to remember.

Fact 2: The opcode indices ai, aj, bi, bj never change because neither x.a nor x.b changes.

Plain lines of the result can be built up by copying lines from x.b to x.results:

'replace'   x.results.extend(x.b[b1:b2])
'delete'    do nothing  (b1 == b2)
'insert'    x.results.extend(x.b[b1:b2])
'equal'     x.results.extend(x.b[b1:b2])

Fact 3: The opcodes cover both x.a and x.b, in order, without any gaps.

This is an explicit requirement of sm.get_opcode:

  • The first tuple has ai==aj==bi==bj==0.

  • Remaining tuples have ai == (aj from the preceding tuple) and bi == (bj from the previous tuple).

Keep in mind this crucial picture:

  • The slices x.a[ai:aj] cover the x.a array, in order without gaps.

  • The slices x.b[bi:bj] cover the x.b array, in order without gaps.

Aha: the x.sentinels array

Mulder’s original algorithm was hard to understand or to change. The culprit was the x.mapping array, which mapped indices into arrays of lines with sentinels to indices into arrays of lines without sentinels.

The new algorithm replaces the x.mapping array with the x.sentinels array. As a result, diff indices never need to be adjusted and handling diff opcodes is easy.

For any index i, x.sentinels[i] is the (possibly empty) list of sentinel lines that precede line a[i]. Computing x.sentinels from old_private_lines is easy. Crucially, x.a and x.sentinels are parallel arrays. That is, len(x.a) == len(x.sentinels), so indices into x.a are also indices into x.sentinels.

Strategy & proof of correctness

Given the x.sentinels array, the strategy for creating the results is simple. Given indices ai, aj, bi, bj from an opcode, the algorithm:

  • Writes sentinels from x.sentinels[i], for i in range(ai,aj).

  • Writes plain lines from b[i], for i in range(bi,bj).

This “just works” because the indices cover both a and b.

  • The algorithm writes sentinels exactly once (in order) because each sentinel appears in x.sentinels[i] for some i in range(len(x.a)).

  • The algorithm writes plain lines exactly once (in order) because each plain line appears in x.b[i] for some i in range(len(x.b)).

This completes an informal proof of the correctness of the algorithm.

The leading and trailing sentinels lines are easy special cases. This code, appearing before the main loop, ensures that leading lines are written first, and only once:

code-block:

x.put_sentinels(0)
x.sentinels[0] = []

Similarly, this line, at the end of the main loop, writes trailing sentinels:

code-block:

x.results.extend(x.trailing_sentinels)

Summary

The algorithm creates an updated set of lines with sentinels using the @clean outline and the updated external file. These new lines then replace the original @clean with a new @clean tree. The algorithm uses only difflib. It will work with any kind of text file. No knowledge of any language is needed.

The algorithm depends on simple, guaranteed, properties of indices in SequenceMatcher opcodes.

The algorithm steps through x.sentinels and x.b, extending x.results as it goes.

The algorithm gets all needed data directly from opcode indices into x.sentinels and x.b. Using opcode indices requires neither reader classes nor auxiliary indices.

The algorithm is simple enough to be understood at first reading. I’ll remember its details for the rest of my life.

Why I like Python

I wrote this soon after discovering Python in 2001. The conclusions are still valid today.

I’ve known for a while that Python was interesting; I attended a Python conference last year and added Python support to Leo. But last week I got that Python is something truly remarkable. I wanted to convert Leo from wxWindows to wxPython, so I began work on c2py, a Python script that would help convert from C++ syntax to Python. While doing so, I had an Aha experience. Python is more than an incremental improvement over Smalltalk or C++ or objective-C; it is “something completely different”. The rest of this post tries to explain this difference.

Clarity

What struck me first as I converted C++ code to Python is how much less blah, blah, blah there is in Python. No braces, no stupid semicolons and most importantly, no declarations. No more pointless distinctions between const, char *, char const *, char * and wxString. No more wondering whether a variable should be signed, unsigned, short or long.

Declarations add clutter, declarations are never obviously right and declarations don’t prevent memory allocation tragedies. Declarations also hinder prototyping. In C++, if I change the type of something I must change all related declarations; this can be a huge and dangerous task. With Python, I can change the type of an object without changing the code at all! It’s no accident that Leo’s new log pane was created first in Python.

Functions returning tuples are a “minor” feature with a huge impact on code clarity. No more passing pointers to data, no more defining (and allocating and deallocating) temporary structs to hold multiple values.

Python can’t check declarations because there aren’t any. However, there is a really nifty tool called pylint that does many of the checks typically done by compilers.

Power

Python is much more powerful than C++, not because Python has more features, but because Python needs less features. Some examples:

  • Python does everything that the C++ Standard Template Library (STL) does, without any of the blah, blah, blah needed by STL. No fuss, no muss, no code bloat.

  • Python’s slicing mechanism is very powerful and applies to any sequence (string, list or tuple). Python’s string library does more with far less functions because slices replace many functions typically found in other string libraries.

  • Writing dict = {} creates a dictionary (hash table). Hash tables can contain anything, including lists and other hash tables.

  • Python’s special functions, __init__, __del__, __repr__, __cmp__, etc. are an elegant way to handle any special need that might arise.

Safety

Before using Python I never fully realized how difficult and dangerous memory allocation is in C++. Try doing:

code-block:

aList[i:j] = list(aString)

in C. You will write about 20 lines of C code. Any error in this code will create a memory allocation crash or leak.

Python is fundamentally safe. C++ is fundamentally unsafe. When I am using Python I am free from worry and anxiety. When I am using C++ I must be constantly “on guard.” A momentary lapse can create a hard-to-find pointer bug. With Python, almost nothing serious can ever go wrong, so I can work late at night, or after a beer. The Python debugger is always available. If an exception occurs, the debugger/interpreter tells me just what went wrong. I don’t have to plan a debugging strategy! Finally, Python recovers from exceptions, so Leo can keep right on going even after a crash!

Speed

Python has almost all the speed of C. Other interpretive environments such as icon and Smalltalk have clarity, power and safety similar to Python. What makes Python unique is its seamless way of making C code look like Python code. Python executes at essentially the speed of C code because most Python modules are written in C. The overhead in calling such modules is negligible. Moreover, if code is too slow, one can always create a C module to do the job.

In fact, Python encourages optimization by moving to higher levels of expression. For example, Leo’s Open command reads an XML file. If this command is too slow I can use Python’s XML parser module. This will speed up Leo while at the same time raising the level of the code.

Conclusions

Little of Python is completely new. What stands out is the superb engineering judgment evident in Python’s design. Python is extremely powerful, yet small, simple and elegant. Python allows me to express my intentions clearly and at the highest possible level.

The only hope of making Leo all it can be is to use the best possible tools. I believe Python will allow me to add, at long last, the new features that Leo should have.

Edward K. Ream, October 25, 2001. P.S., September, 2005:

Four years of experience have only added to my admiration for Python. Leo could not possibly be what it is today without Python.