diff --git a/kibot/mcpyrate/__init__.py b/kibot/mcpyrate/__init__.py
index f677b11d..0f20a494 100644
--- a/kibot/mcpyrate/__init__.py
+++ b/kibot/mcpyrate/__init__.py
@@ -1,3 +1,5 @@
+"""mcpyrate: Advanced macro expander and language lab for Python."""
+
 from .astdumper import dump  # noqa: F401
 from .core import MacroExpansionError  # noqa: F401
 from .dialects import Dialect  # noqa: F401
@@ -5,4 +7,12 @@ from .expander import namemacro, parametricmacro  # noqa: F401
 from .unparser import unparse  # noqa: F401
 from .utils import gensym  # noqa: F401
 
-__version__ = "3.5.4"
+# For public API inspection, import modules that wouldn't otherwise get imported.
+from . import ansi  # noqa: F401
+from . import debug  # noqa: F401
+from . import metatools  # noqa: F401
+from . import pycachecleaner  # noqa: F401
+from . import quotes  # noqa: F401
+from . import splicing  # noqa: F401
+
+__version__ = "3.6.0"
diff --git a/kibot/mcpyrate/coreutils.py b/kibot/mcpyrate/coreutils.py
index 824f71be..248abf88 100644
--- a/kibot/mcpyrate/coreutils.py
+++ b/kibot/mcpyrate/coreutils.py
@@ -2,7 +2,8 @@
 """Utilities related to writing macro expanders and similar meta-metaprogramming tasks."""
 
 __all__ = ["resolve_package", "relativize", "match_syspath",
-           "ismacroimport", "get_macros"]
+           "ismacroimport", "get_macros",
+           "isfutureimport", "split_futureimports", "inject_after_futureimports"]
 
 import ast
 import importlib
@@ -12,7 +13,7 @@ import pathlib
 import sys
 
 from .unparser import unparse_with_fallbacks
-from .utils import format_location
+from .utils import format_location, getdocstring
 
 
 def resolve_package(filename):  # TODO: for now, `guess_package`, really. Check the docs again.
@@ -90,6 +91,15 @@ def ismacroimport(statement, magicname='macros'):
 def get_macros(macroimport, *, filename, reload=False, allow_asname=True, self_module=None):
     """Get absolute module name, macro names and macro functions from a macro-import.
 
+    A macro-import is a statement of the form::
+
+        from ... import macros, ...
+
+    where `macros` is the magic name that your actual macro expander uses to recognize
+    a macro-import (see `ismacroimport`). This function does not care about what the
+    actual magic name is, and simply ignores the first name that is imported by the
+    import statement.
+
     As a side effect, import the macro definition module.
 
     Return value is `module_absname, {macroname0: macrofunction0, ...}`.
@@ -166,7 +176,7 @@ def get_macros(macroimport, *, filename, reload=False, allow_asname=True, self_m
             module = importlib.reload(module)
 
     bindings = {}
-    for name in macroimport.names[1:]:
+    for name in macroimport.names[1:]:  # skip the "macros" in `from ... import macros, ...`
         if not allow_asname and name.asname is not None:
             approx_sourcecode = unparse_with_fallbacks(macroimport, debug=True, color=True)
             loc = format_location(filename, macroimport, approx_sourcecode)
@@ -241,3 +251,54 @@ def _mcpyrate_attr(dotted_name, *, force_import=False):
         value = ast.Attribute(value=value, attr=name)
 
     return value
+
+# --------------------------------------------------------------------------------
+
+def isfutureimport(tree):
+    """Return whether `tree` is a `from __future__ import ...`."""
+    return isinstance(tree, ast.ImportFrom) and tree.module == "__future__"
+
+def split_futureimports(body):
+    """Split `body` into `__future__` imports and everything else.
+
+    `body`: list of `ast.stmt`, the suite representing a module top level.
+
+    Return value is `[docstring, future_imports, the_rest]`, where each item
+    is a list of `ast.stmt` (possibly empty).
+    """
+    if getdocstring(body):
+        docstring, *body = body
+        docstring = [docstring]
+    else:
+        docstring = []
+
+    k = -1  # ensure `k` gets defined even if `body` is empty
+    for k, bstmt in enumerate(body):
+        if not isfutureimport(bstmt):
+            break
+    if k >= 0:
+        return docstring, body[:k], body[k:]
+    return docstring, [], body
+
+def inject_after_futureimports(stmts, body):
+    """Inject one or more statements into `body` after its `__future__` imports.
+
+    `stmts`: `ast.stmt` or list of `ast.stmt`, the statement(s) to inject.
+    `body`: list of `ast.stmt`, the suite representing a module top level.
+
+    Return value is the list `[docstring] + futureimports + stmts + the_rest`.
+
+    If `body` has no docstring node at its beginning, the docstring part is
+    automatically omitted.
+
+    If `body` has no `__future__` imports at the beginning just after the
+    optional docstring, the `futureimports` part is automatically omitted.
+    """
+    if not isinstance(body, list):
+        raise TypeError(f"`body` must be a `list`, got {type(body)} with value {repr(body)}")
+    if not isinstance(stmts, list):
+        if not isinstance(stmts, ast.stmt):
+            raise TypeError(f"`stmts` must be `ast.stmt` or a `list` of `ast.stmt`, got {type(stmts)} with value {repr(stmts)}")
+        stmts = [stmts]
+    docstring, futureimports, body = split_futureimports(body)
+    return docstring + futureimports + stmts + body
diff --git a/kibot/mcpyrate/dialects.py b/kibot/mcpyrate/dialects.py
index 8dd69da5..0f53c46a 100644
--- a/kibot/mcpyrate/dialects.py
+++ b/kibot/mcpyrate/dialects.py
@@ -15,10 +15,22 @@ from .utils import format_macrofunction
 
 
 class Dialect:
-    """Base class for dialects."""
+    """Base class for dialects.
+
+    `expander`: the `DialectExpander` instance. The expander provides this automatically.
+                Stored as `self.expander`.
+
+    During dialect expansion, the source location info of the dialect-import statement
+    that invoked this dialect-import is available as `self.lineno` and `self.col_offset`.
+
+    You can pass those to `mcpyrate.splicing.splice_dialect` to automatically mark the
+    lines from your dialect template as coming from that dialect-import in the user
+    source code.
+    """
     def __init__(self, expander):
-        """`expander`: the `DialectExpander` instance. The expander provides this automatically."""
         self.expander = expander
+        self.lineno = None
+        self.col_offset = None
 
     def transform_source(self, text):
         """Override this to add a whole-module source transformer to your dialect.
@@ -27,10 +39,10 @@ class Dialect:
         tells the expander this dialect does not provide a source transformer.
 
         Rarely needed. Because we don't (yet) have a generic, extensible
-        tokenizer for "Python-plus" with extended surface syntax, this is
-        currently essentially a per-module hook to plug in a transpiler
-        that compiles source code from some other programming language
-        into macro-enabled Python.
+        tokenizer for "Python-plus" with extended surface syntax, not to mention that
+        none of the available Python dev tools support any such, this is currently
+        essentially a per-module hook to plug in a transpiler that compiles
+        source code from some other programming language into macro-enabled Python.
 
         The dialect system autodetects the text encoding the same way Python itself
         does. That is, it reads the magic comment at the top of the source file
@@ -80,8 +92,8 @@ class Dialect:
         Output should be the transformed AST.
 
         To easily splice `tree.body` into your template, see the utility
-        `mcpyrate.splicing.splice_dialect` (it automatically handles macro-imports,
-        dialect-imports, the magic `__all__`, and the module docstring).
+        `mcpyrate.splicing.splice_dialect` (it automatically handles future-imports,
+        macro-imports, dialect-imports, the magic `__all__`, and the module docstring).
 
         As an example, see the `dialects` module in `unpythonic` for example dialects.
 
@@ -159,7 +171,7 @@ class DialectExpander:
 
     Due to modularity requirements introduced by `mcpyrate`'s support for
     multi-phase compilation (see the module `mcpyrate.multiphase`), this
-    class is a bit technical to use. See `mcpyrate.importer`. Roughly,
+    class is a bit technical to use. See `mcpyrate.compiler`. Roughly,
     for a single-phase compile::
 
         dexpander = DialectExpander(filename=...)
@@ -221,11 +233,11 @@ class DialectExpander:
         # state is so last decade.
         dialect_instances = []
         while True:
-            module_absname, bindings = find_dialectimport(content)
-            if not module_absname:  # no more dialects
+            theimport = find_dialectimport(content)
+            if theimport:
+                module_absname, bindings, lineno, col_offset = theimport
+            else:  # no more dialects
                 break
-            if not bindings:
-                continue
 
             for dialectname, cls in bindings.items():
                 if not (isinstance(cls, type) and issubclass(cls, Dialect)):
@@ -235,6 +247,9 @@ class DialectExpander:
                     dialect = cls(expander=self)
                 except Exception as err:
                     raise ImportError(f"Unexpected exception while instantiating dialect `{module_absname}.{dialectname}`") from err
+                # make the dialect-import source location info available to the transformers
+                dialect.lineno = lineno
+                dialect.col_offset = col_offset
 
                 try:
                     transformer_method = getattr(dialect, transform)
@@ -338,9 +353,24 @@ class DialectExpander:
         So we can only rely on the literal text "from ... import dialects, ...",
         similarly to how Racket heavily constrains the format of its `#lang` line.
 
-        Return value is a dict `{dialectname: class, ...}` with all collected bindings
-        from that one dialect-import. Each binding is a dialect, so usually there is
-        just one.
+        Return value is the tuple `(module_absname, bindings, lineno, col_offset)`:
+
+            - `module_absname` is the absolute module name referred to by the import
+            - `bindings` is a dict `{dialectname: class, ...}`, with all bindings
+              collected from that one dialect-import statement. Each binding is a
+              dialect, so usually there is just one.
+            - `lineno` is the line number of the import statement, determined by
+              counting the lines of `text`.
+            - `col_offset` is the corresponding column offset.
+              Currently not extracted; is always set to 0.
+
+        The return value refers to the first not-yet-seen dialect-import (according
+        to the private cache `self._seen`). Note that this does not transform away
+        the dialect-imports, because the expander still needs to see them in the
+        AST transformation step.
+
+        If there are no more dialect-imports that have not been seen already,
+        the return value is `None`.
         """
         matches = _dialectimport.finditer(text)
         try:
@@ -349,15 +379,17 @@ class DialectExpander:
                 statement = match.group(0).strip()
                 if statement not in self._seen:  # apply each unique dialect-import once
                     self._seen.add(statement)
+                    lineno = 1 + text[0:match.start()].count("\n")  # https://stackoverflow.com/a/48647994
+                    col_offset = 0  # TODO: extract the correct column offset
                     break
         except StopIteration:
-            return "", {}
+            return None
 
         dummy_module = ast.parse(statement, filename=self.filename, mode="exec")
         dialectimport = dummy_module.body[0]
         module_absname, bindings = get_macros(dialectimport, filename=self.filename,
                                               reload=False, allow_asname=False)
-        return module_absname, bindings
+        return module_absname, bindings, lineno, col_offset
 
     def find_dialectimport_ast(self, tree):
         """Find the first dialect-import statement by scanning the AST `tree`.
@@ -374,20 +406,38 @@ class DialectExpander:
 
             from ... import dialects, ...
 
-        Return value is a dict `{dialectname: class, ...}` with all collected bindings
-        from that one dialect-import. Each binding is a dialect, so usually there is
-        just one.
+        Return value is the tuple `(module_absname, bindings, lineno)`, where:
+
+            - `module_absname` is the absolute module name referred to by the import
+            - `bindings` is a dict `{dialectname: class, ...}`, with all bindings
+              collected from that one dialect-import statement. Each binding is a
+              dialect, so usually there is just one.
+            - `lineno` is the line number from the import statement node,
+              or `None` if the statement had no `lineno` attribute.
+            - `col_offset` is the corresponding column offset.
+              It is also taken from the same import statement node.
+
+        The return value refers to the first dialect-import that has not yet been
+        transformed away. If there are no more dialect-imports, the return value
+        is `None`.
         """
         for index, statement in enumerate(tree.body):
             if ismacroimport(statement, magicname="dialects"):
                 break
         else:
-            return "", {}
+            return None
 
         module_absname, bindings = get_macros(statement, filename=self.filename,
                                               reload=False, allow_asname=False)
         # Remove all names to prevent dialects being used as regular run-time objects.
         # Always use an absolute import, for the unhygienic expose API guarantee.
-        tree.body[index] = ast.copy_location(ast.Import(names=[ast.alias(name=module_absname, asname=None)]),
+        thealias = ast.copy_location(ast.alias(name=module_absname, asname=None),
+                                     statement)
+        tree.body[index] = ast.copy_location(ast.Import(names=[thealias]),
                                              statement)
-        return module_absname, bindings
+
+        # Get source location info
+        lineno = statement.lineno if hasattr(statement, "lineno") else None
+        col_offset = statement.col_offset if hasattr(statement, "col_offset") else None
+
+        return module_absname, bindings, lineno, col_offset
diff --git a/kibot/mcpyrate/expander.py b/kibot/mcpyrate/expander.py
index d4c081db..ebd6727e 100644
--- a/kibot/mcpyrate/expander.py
+++ b/kibot/mcpyrate/expander.py
@@ -649,12 +649,10 @@ def find_macros(tree, *, filename, reload=False, self_module=None, transform=Tru
                 else:
                     # Remove all names to prevent macros being used as regular run-time objects.
                     # Always use an absolute import, for the unhygienic expose API guarantee.
-                    tree.body[index] = copy_location(Import(names=[
-                        alias(name=module_absname,
-                              asname=None,
-                              lineno=getattr(statement, 'lineno', 0),
-                              col_offset=getattr(statement, 'col_offset', 0))]),
-                        statement)
+                    thealias = copy_location(alias(name=module_absname, asname=None),
+                                             statement)
+                    tree.body[index] = copy_location(Import(names=[thealias]),
+                                                     statement)
     for index in reversed(stmts_to_delete):
         tree.body.pop(index)
     return bindings
diff --git a/kibot/mcpyrate/importer.py b/kibot/mcpyrate/importer.py
index 230381be..4e58e963 100644
--- a/kibot/mcpyrate/importer.py
+++ b/kibot/mcpyrate/importer.py
@@ -103,6 +103,7 @@ def path_stats(path, _stats_cache=None):
 
     Beside the source file `path` itself, we look at any macro definition files
     the source file imports macros from, recursively, in a `make`-like fashion.
+    Dialect-imports, if any, are treated the same way.
 
     `_stats_cache` is used internally to speed up the computation, in case the
     dependency graph hits the same source file multiple times.
diff --git a/kibot/mcpyrate/metatools.py b/kibot/mcpyrate/metatools.py
index 9ae4711f..1eddfce3 100644
--- a/kibot/mcpyrate/metatools.py
+++ b/kibot/mcpyrate/metatools.py
@@ -38,7 +38,8 @@ __all__ = ["macro_bindings",
            "expand1s", "expands",
            "expand1rq", "expandrq",
            "expand1r", "expandr",
-           "stepr"]
+           "stepr",
+           "expand_first"]
 
 import ast
 
@@ -48,6 +49,8 @@ from .coreutils import _mcpyrate_attr
 from .debug import step_expansion  # noqa: F401, used in macro output.
 from .expander import MacroExpander, namemacro, parametricmacro
 from .quotes import astify, capture_value, q, unastify
+from .unparser import unparse_with_fallbacks
+from .utils import extract_bindings
 
 
 def _mcpyrate_metatools_attr(attr):
@@ -447,3 +450,78 @@ def stepr(tree, *, args, syntax, expander, **kw):
                     [ast.keyword("args", astify(args)),
                      ast.keyword("syntax", astify(syntax)),
                      ast.keyword("expander", expander_node)])
+
+# --------------------------------------------------------------------------------
+
+@parametricmacro
+def expand_first(tree, *, args, syntax, expander, **kw):
+    """[syntax, block] Force given macros to expand before other macros.
+
+    Usage::
+
+        with expand_first[macro0, ...]:
+            ...
+
+    Each argument can be either a bare macro name, e.g. `macro0`, or a
+    hygienically unquoted macro, e.g. `q[h[macro0]]`.
+
+    As an example, consider::
+
+        with your_block_macro:
+            macro0[expr]
+
+    In this case, if `your_block_macro` expands outside-in, it will transform the
+    `expr` inside the `macro0[expr] before `macro0` even sees the AST. If the test
+    fails or errors, the error message will contain the expanded version of `expr`,
+    not the original one. Now, if we change the example to::
+
+        with expand_first[macro0]:
+            with your_block_macro:
+                macro0[expr]
+
+    In this case, `expand_first` arranges things so that `macro0[expr]` expands first
+    (even if `your_block_macro` expands outside-in), so it will see the original,
+    unexpanded AST.
+
+    This does imply that `your_block_macro` will then receive the expanded form of
+    `macro0[expr]` as input, but that's macros for you.
+
+    There is no particular ordering in which the given set of macros expands;
+    they are handled by one expander with bindings registered for all of them.
+    (So the expansion order is determined by the order their use sites are
+     encountered; the ordering of the names in the argument list does not matter.)
+    """
+    if syntax != "block":
+        raise SyntaxError("expand_first is a block macro only")  # pragma: no cover
+    if syntax == "block" and kw['optional_vars'] is not None:
+        raise SyntaxError("expand_first does not take an as-part")  # pragma: no cover
+    if not args:
+        raise SyntaxError("expected a comma-separated list of `macroname` or `q[h[macroname]]` in `with expand_first[macro0, ...]:`; no macro arguments were given")
+
+    # Expand macros in `args` to handle `q[h[somemacro]]`.
+    #
+    # We must use the full `bindings`, not only those of the quasiquote operators
+    # (see `mcpyrate.quotes._expand_quasiquotes`), so that the expander recognizes
+    # which hygienic captures are macros.
+    args = MacroExpander(expander.bindings, filename=expander.filename).visit(args)
+
+    # In the arguments, we should now have `Name` nodes only.
+    invalid_args = [node for node in args if type(node) is not ast.Name]
+    if invalid_args:
+        invalid_args_str = ", ".join(unparse_with_fallbacks(node, color=True, debug=True)
+                                     for node in invalid_args)
+        raise SyntaxError(f"expected a comma-separated list of `macroname` or `q[h[macroname]]` in `with expand_first[macro0, ...]:`; invalid args: {invalid_args_str}")
+
+    # Furthermore, all the specified names must be bound as macros in the current expander.
+    invalid_args = [name_node for name_node in args if name_node.id not in expander.bindings]
+    if invalid_args:
+        invalid_args_str = ", ".join(unparse_with_fallbacks(node, color=True, debug=True)
+                                     for node in invalid_args)
+        raise SyntaxError(f"all macro names in `with expand_first[macro0, ...]:` must be bound in the current expander; the following are not: {invalid_args_str}")
+
+    # All ok. First map the names to macro functions:
+    macros = [expander.bindings[name_node.id] for name_node in args]
+
+    # Then map the macro functions to *all* their names in `expander`:
+    macro_bindings = extract_bindings(expander.bindings, *macros)
+    return MacroExpander(macro_bindings, filename=expander.filename).visit(tree)
diff --git a/kibot/mcpyrate/multiphase.py b/kibot/mcpyrate/multiphase.py
index fbb73d61..a68ab5e4 100644
--- a/kibot/mcpyrate/multiphase.py
+++ b/kibot/mcpyrate/multiphase.py
@@ -26,10 +26,9 @@ from copy import copy, deepcopy
 
 from . import compiler
 from .colorizer import ColorScheme, setcolor
-from .coreutils import ismacroimport
+from .coreutils import ismacroimport, isfutureimport, inject_after_futureimports
 from .expander import destructure_candidate, namemacro, parametricmacro
 from .unparser import unparse_with_fallbacks
-from .utils import getdocstring
 
 # --------------------------------------------------------------------------------
 # Private utilities.
@@ -71,10 +70,6 @@ def iswithphase(stmt, *, filename):
 
     return n
 
-def isfutureimport(tree):
-    """Return whether `tree` is a `from __future__ import ...`."""
-    return isinstance(tree, ast.ImportFrom) and tree.module == "__future__"
-
 def extract_phase(tree, *, filename, phase=0):
     """Split `tree` into given `phase` and remaining parts.
 
@@ -142,35 +137,6 @@ def extract_phase(tree, *, filename, phase=0):
     newmodule.body = thisphase
     return newmodule
 
-def split_futureimports(body):
-    """Split `body` into `__future__` imports and everything else.
-
-    `body`: list of `ast.stmt`, the suite representing a module top level.
-
-    Returns `[future_imports, the_rest]`.
-    """
-    k = -1  # ensure `k` gets defined even if `body` is empty
-    for k, bstmt in enumerate(body):
-        if not isfutureimport(bstmt):
-            break
-    if k >= 0:
-        return body[:k], body[k:]
-    return [], body
-
-def inject_after_futureimports(stmt, body):
-    """Inject a statement into `body` after `__future__` imports.
-
-    `body`: list of `ast.stmt`, the suite representing a module top level.
-    `stmt`: `ast.stmt`, the statement to inject.
-    """
-    if getdocstring(body):
-        docstring, *body = body
-        futureimports, body = split_futureimports(body)
-        return [docstring] + futureimports + [stmt] + body
-    else:  # no docstring
-        futureimports, body = split_futureimports(body)
-        return futureimports + [stmt] + body
-
 # --------------------------------------------------------------------------------
 # Public utilities.
 
diff --git a/kibot/mcpyrate/quotes.py b/kibot/mcpyrate/quotes.py
index 8e1278e8..504b9285 100644
--- a/kibot/mcpyrate/quotes.py
+++ b/kibot/mcpyrate/quotes.py
@@ -536,7 +536,7 @@ def is_captured_macro(tree):
            the next one thousand years"; see `mcpyrate.gensym`, which links to
            the UUID spec used by the implementation.)
 
-        - `frozen_macro` is either `bytes` object that stores a reference to the
+        - `frozen_macro` is a `bytes` object that stores a reference to the
           frozen macro function as opaque binary data.
 
         The `bytes` object can be decoded by passing the whole return value as `key`
diff --git a/kibot/mcpyrate/splicing.py b/kibot/mcpyrate/splicing.py
index 06c28fe5..7ff3e197 100644
--- a/kibot/mcpyrate/splicing.py
+++ b/kibot/mcpyrate/splicing.py
@@ -7,7 +7,7 @@ import ast
 from copy import deepcopy
 
 from .astfixers import fix_locations
-from .coreutils import ismacroimport
+from .coreutils import ismacroimport, split_futureimports
 from .markers import ASTMarker
 from .utils import getdocstring
 from .walkers import ASTTransformer
@@ -157,29 +157,64 @@ def splice_statements(body, template, tag="__paste_here__"):
     return StatementSplicer().visit(template)
 
 
-def splice_dialect(body, template, tag="__paste_here__"):
+def splice_dialect(body, template, tag="__paste_here__", lineno=None, col_offset=None):
     """In a dialect AST transformer, splice module `body` into `template`.
 
     On top of what `splice_statements` does, this function handles macro-imports
     and dialect-imports specially, gathering them all at the top level of the
     final module body, so that mcpyrate sees them when the module is sent to
-    the macro expander.
+    the macro expander. This is to allow a dialect template to splice the body
+    into the inside of a `with` block (e.g. to invoke some code-walking macro
+    that changes the language semantics, such as an auto-TCO or a lazifier),
+    without breaking macro-imports (and further dialect-imports) introduced
+    by user code in the body.
 
     Any dialect-imports in the template are placed first (in the order they
     appear in the template), followed by any dialect-imports in the user code
     (in the order they appear in the user code), followed by macro-imports in
     the template, then macro-imports in the user code.
 
-    This also handles the module docstring and the magic `__all__` (if any)
-    from `body`. The docstring comes first, before dialect-imports. The magic
-    `__all__` is placed after dialect-imports, before macro-imports.
+    We also handle the module docstring, future-imports, and the magic `__all__`.
+
+    The optional `lineno` and `col_offset` parameters can be used to tell
+    `splice_dialect` the source location info of the dialect-import (in the
+    unexpanded source code) that triggered this template. If specified, they
+    are used to mark all the lines coming from the template as having come
+    from that dialect-import statement. During dialect expansion, you can
+    get these from the `lineno` and `col_offset` attributes of your dialect
+    instance (these attributes are filled in by `DialectExpander`).
+
+    If both `body` and `template` have a module docstring, they are concatenated
+    to produce the module docstring for the result. If only one of them has a
+    module docstring, that docstring is used as-is. If neither has a module docstring,
+    the docstring is omitted.
+
+    The primary use of a module docstring in a dialect template is to be able to say
+    that the program was written in dialect X, more information on which can be found at...
+
+    Future-imports from `template` and `body` are concatenated.
+
+    The magic `__all__` is taken from `body`; if `body` does not define it,
+    it is omitted.
+
+    In the result, the ordering is::
+
+        docstring
+        template future-imports
+        body future-imports
+        __all__ (if defined in body)
+        template dialect-imports
+        body dialect-imports
+        template macro-imports
+        body macro-imports
+        the rest
 
     Parameters:
 
-        `body`: `list` of statements
+        `body`: `list` of `ast.stmt`, or a single `ast.stmt`
             Original module body from the user code (input).
 
-        `template`: `list` of statements
+        `template`: `list` of `ast.stmt`, or a single `ast.stmt`
             Template for the final module body (output).
 
             Must contain a paste-here indicator as in `splice_statements`.
@@ -187,16 +222,19 @@ def splice_dialect(body, template, tag="__paste_here__"):
         `tag`: `str`
             The name of the paste-here indicator in `template`.
 
-    Returns `template` with `body` spliced in. Note `template` is **not** copied,
-    and will be mutated in-place.
+        `lineno`: optional `int`
+        `col_offset`: optional `int`
+            Source location info of the dialect-import that triggered this template.
 
-    Also `body` is mutated, to remove macro-imports, `__all__` and the module
-    docstring; these are pasted into the final result.
+    Return value is `template` with `body` spliced in.
+
+    Note `template` and `body` are **not** copied, and **both** will be mutated
+    during the splicing process.
     """
     if isinstance(body, ast.AST):
         body = [body]
     if isinstance(template, ast.AST):
-        body = [template]
+        template = [template]
     if not body:
         raise ValueError("expected at least one statement in `body`")
     if not template:
@@ -207,20 +245,34 @@ def splice_dialect(body, template, tag="__paste_here__"):
     # Even if they have location info, it's for a different file compared
     # to the use site where `body` comes from.
     #
-    # Pretend the template code appears at the beginning of the user module.
-    #
-    # TODO: It would be better to pretend it appears at the line that has the dialect-import.
-    # TODO: Requires a `lineno` parameter here, and `DialectExpander` must be modified to supply it.
-    # TODO: We could extract the `lineno` in `find_dialectimport_ast` and then pass it to the
-    # TODO: user-defined dialect AST transformer, so it could pass it to us if it wants to.
-    for stmt in template:
-        fix_locations(stmt, body[0], mode="overwrite")
-
-    if getdocstring(body):
-        docstring, *body = body
-        docstring = [docstring]
+    # Pretend the template code appears at the given source location,
+    # or if not given, at the beginning of `body`.
+    if lineno is not None and col_offset is not None:
+        srcloc_dummynode = ast.Constant(value=None)
+        srcloc_dummynode.lineno = lineno
+        srcloc_dummynode.col_offset = col_offset
     else:
-        docstring = []
+        srcloc_dummynode = body[0]
+    for stmt in template:
+        fix_locations(stmt, srcloc_dummynode, mode="overwrite")
+
+    user_docstring, user_futureimports, body = split_futureimports(body)
+    template_docstring, template_futureimports, template = split_futureimports(template)
+
+    # Combine user and template docstrings if both are defined.
+    if user_docstring and template_docstring:
+        # We must extract the bare strings, combine them, and then pack the result into an AST node.
+        user_doc = getdocstring(user_docstring)
+        template_doc = getdocstring(template_docstring)
+        sep = "\n" + ("-" * 79) + "\n"
+        new_doc = user_doc + sep + template_doc
+        new_docstring = ast.copy_location(ast.Constant(value=new_doc),
+                                          user_docstring[0])
+        docstring = [new_docstring]
+    else:
+        docstring = user_docstring or template_docstring
+
+    futureimports = template_futureimports + user_futureimports
 
     def extract_magic_all(tree):
         def ismagicall(tree):
@@ -239,6 +291,7 @@ def splice_dialect(body, template, tag="__paste_here__"):
         w.visit(tree)
         return tree, w.collected
     body, user_magic_all = extract_magic_all(body)
+    template, ignored_template_magic_all = extract_magic_all(template)
 
     def extract_macroimports(tree, *, magicname="macros"):
         class MacroImportExtractor(ASTTransformer):
@@ -257,6 +310,7 @@ def splice_dialect(body, template, tag="__paste_here__"):
 
     finalbody = splice_statements(body, template, tag)
     return (docstring +
+            futureimports +
             user_magic_all +
             template_dialect_imports + user_dialect_imports +
             template_macro_imports + user_macro_imports +
diff --git a/kibot/mcpyrate/utils.py b/kibot/mcpyrate/utils.py
index 688cf91d..504feb9d 100644
--- a/kibot/mcpyrate/utils.py
+++ b/kibot/mcpyrate/utils.py
@@ -2,7 +2,7 @@
 """General utilities. Can be useful for writing both macros as well as macro expanders."""
 
 __all__ = ["gensym", "scrub_uuid", "flatten", "rename", "extract_bindings", "getdocstring",
-           "format_location", "format_macrofunction", "format_context",
+           "get_lineno", "format_location", "format_macrofunction", "format_context",
            "NestingLevelTracker"]
 
 import ast
@@ -188,34 +188,66 @@ def getdocstring(body):
 
 # --------------------------------------------------------------------------------
 
+def get_lineno(tree):
+    """Extract the source line number from `tree`.
+
+    `tree`: AST node, list of AST nodes, or an AST marker.
+
+    `tree` is searched recursively (depth first) until a `lineno` attribute is found;
+    its value is then returned.
+
+    If no `lineno` attribute is found anywhere inside `tree`, the return value is `None`.
+    """
+    if hasattr(tree, "lineno"):
+        return tree.lineno
+    elif isinstance(tree, markers.ASTMarker) and hasattr(tree, "body"):  # look inside AST markers
+        return get_lineno(tree.body)
+    elif isinstance(tree, ast.AST):  # look inside AST nodes
+        # Note `iter_fields` ignores attribute fields such as line numbers and column offsets,
+        # so we don't recurse into those.
+        for fieldname, node in ast.iter_fields(tree):
+            lineno = get_lineno(node)
+            if lineno:
+                return lineno
+    elif isinstance(tree, list):  # look inside statement suites
+        for node in tree:
+            lineno = get_lineno(node)
+            if lineno:
+                return lineno
+    return None
+
+
 def format_location(filename, tree, sourcecode):
     """Format a source code location in a standard way, for error messages.
 
     `filename`: full path to `.py` file.
-    `tree`: AST node to get source line number from. (Looks inside AST markers.)
+    `tree`: AST node to get source line number from. (Looks inside automatically if needed.)
     `sourcecode`: source code (typically, to get this, `unparse(tree)`
                   before expanding it), or `None` to omit it.
-    """
-    lineno = None
-    if hasattr(tree, "lineno"):
-        lineno = tree.lineno
-    elif isinstance(tree, markers.ASTMarker) and hasattr(tree, "body"):
-        if hasattr(tree.body, "lineno"):
-            lineno = tree.body.lineno
-        elif isinstance(tree.body, list) and tree.body and hasattr(tree.body[0], "lineno"):  # e.g. `SpliceNodes`
-            lineno = tree.body[0].lineno
 
+    Return value is an `str` containing colored text, suitable for terminal output.
+    Example outputs for single-line and multiline source code::
+
+        /path/to/hello.py:42: print("hello!")
+
+        /path/to/hello.py:1337:
+        if helloing:
+            print("hello!")
+    """
     if sourcecode:
         sep = " " if "\n" not in sourcecode else "\n"
         source_with_sep = f"{sep}{sourcecode}"
     else:
         source_with_sep = ""
 
-    return f'{colorize(filename, ColorScheme.SOURCEFILENAME)}:{lineno}:{source_with_sep}'
+    return f'{colorize(filename, ColorScheme.SOURCEFILENAME)}:{get_lineno(tree)}:{source_with_sep}'
 
 
 def format_macrofunction(function):
-    """Format the fully qualified name of a macro function, for error messages."""
+    """Format the fully qualified name of a macro function, for error messages.
+
+    Return value is an `str` with the fully qualified name.
+    """
     # Catch broken bindings due to erroneous imports in user code
     # (e.g. accidentally to a module object instead of to a function object)
     if not (hasattr(function, "__module__") and hasattr(function, "__qualname__")):
@@ -226,7 +258,13 @@ def format_macrofunction(function):
 
 
 def format_context(tree, *, n=5):
-    """Format up to the first `n` lines of source code of `tree`."""
+    """Format up to the first `n` lines of source code of `tree`.
+
+    The source code is produced from `tree` by unparsing.
+
+    Return value is an `str` containing colored text with syntax highlighting,
+    suitable for terminal output.
+    """
     code_lines = unparser.unparse_with_fallbacks(tree, debug=True, color=True).split("\n")
     code = "\n".join(code_lines[:n])
     if len(code_lines) > n:
@@ -255,7 +293,17 @@ class NestingLevelTracker:
     value = property(fget=_get_value, doc="The current level. Read-only. Use `set_to` or `change_by` to change.")
 
     def set_to(self, value):
-        """Context manager. Run a section of code with the level set to `value`."""
+        """Context manager. Run a section of code with the level set to `value`.
+
+        Example::
+
+            t = NestingLevelTracker()
+            assert t.value == 0
+            with t.set_to(42):
+                assert t.value == 42
+                ...
+            assert t.value == 0
+        """
         if not isinstance(value, int):
             raise TypeError(f"Expected integer `value`, got {type(value)} with value {repr(value)}")
         if value < 0:
@@ -271,5 +319,18 @@ class NestingLevelTracker:
         return _set_to()
 
     def changed_by(self, delta):
-        """Context manager. Run a section of code with the level incremented by `delta`."""
+        """Context manager. Run a section of code with the level incremented by `delta`.
+
+        Example::
+
+            t = NestingLevelTracker()
+            assert t.value == 0
+            with t.changed_by(+21):
+                assert t.value == 21
+                with t.changed_by(+21):
+                    assert t.value == 42
+                    ...
+                assert t.value == 21
+            assert t.value == 0
+        """
         return self.set_to(self.value + delta)